TRADING BLOC EXPOSURE IN INTERNATIONAL ASSET PRICING: THE CASE OF AFTA, CER AND NAFTA. Chee-Wooi Hooy and Kim-Leng Goh

Transcription

1 Labuan Bulletn OF INTERNATIONAL BUSINESS & FINANCE Labuan Bulletn of Internatonal Busness & Fnance 3, 2005, ISSN TRADING BLOC EXPOSURE IN INTERNATIONAL ASSET PRICING: THE CASE OF AFTA, CER AND NAFTA Chee-Woo Hooy and Km-Leng Goh Faculty of Economcs and Admnstraton, Unversty of Malaya, Malaysa Abstract Ths paper shows that the resurgence of trade regonalsm has a sgnfcant mpact on stock market returns of the member countres n the ASEAN Free Trade Area (AFTA), Australa-New Zealand Closer Economc Relatons Trade Arrangement (CER) and North Amercan Free Trade Area (NAFTA). A tradng bloc nternatonal captal asset prcng model (ICAPM) s proposed and we fnd that the tradng bloc factor ncreases the explanatory power of the conventonal ICAPM for AFTA and CER. Evdence ndcates that returns of the markets n AFTA and CER are hghly exposed to the tradng bloc factor. At the same tme, exposure to the global market s stll sgnfcant, partcularly for the more advanced markets of Sngapore and Australa. The conventonal ICAPM s stll relevant for the large and leadng world markets n NAFTA. The trade bloc factor, however, has mnmal mpact n nfluencng market returns of non-member countres. The fndngs of sgnfcant exposure to regonal dynamcs offer an explanaton to why stock markets are generally segmented. Keywords: APEC; ntegraton; trade bloc; nternatonal CAPM; GARCH. 1. Introducton The 1990s wtnessed the resurgence of nterest n regonal trade agreements n spte of the attempt to promote multlateral trade negotatons n the WTO agenda. Regonal ntegraton has oen been seen as a means to mprove the compettveness of member countres and ntegrate them nto the world multlateral tradng system. The new regonalsm features more than trade ntegraton, and ntatves are taken to deepen regonal fnancal ntegraton through areas such as currency arrangements, monetary polces and fnance. Artcle 1109 n the North Amercan Free Trade Agreement Correspondng author: Km-Leng Goh, Faculty of Economcs and Admnstraton, Unversty of Malaya, Kuala Lumpur, Malaysa. Emal: klgoh@um.edu.my.

2 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, (NAFTA), for example, calls for free and quck transfers of all payments relatng to equty transactons ncludng dvdends, nterest and captal gans among the members. The December 1995 Assocaton of Southeast Asan Natons (ASEAN) Summt endorsed n prncple the concept of an nvestment area to lower and remove barrers to ntra-regonal nvestment among members of the ASEAN Free Trade Area (AFTA) (ASEAN Secretarat, The am of ths paper s to nvestgate the mportance of trade regonalsm n asset prcng. Ths s motvated by the fact that economc ntegraton through formaton of tradng blocs not only promotes real convergence but also actve fnancal ntegraton among member countres (Fratzscher, 2002). The strengthenng of a regonal tradng bloc could stmulate both drect and portfolo nvestment from and for the member countres. As the core segment of the fnancal system, the stock market movements are expected to show more convergence wthn the tradng bloc. When tradng systems are homogenzed through free trade arrangements, the ncreased economc nterdependence among the bloc members could boost captal movements and lead to stock prcng convergence. Furthermore, efforts are lkely to be taken to acheve harmonzaton of captal market regulatons, synchronzaton of monetary polcy, and coordnaton of exchange rate management to stablze the party of compettve advantages among the member states and to reduce arbtrage opportuntes. Exstng lterature on stock market ntegraton offers lmted nsght on the mpact of regonalsm. The closest to ths s the work of Lessard (1973) that nvestgated the opportunty of dversfcaton n an nvestment unon (a concept parallel to custom unon) formed by 4 Latn countres comprsng Argentna, Brazl, Chle and Columba. Usng multvarate factor analyss on data for , the study reported that stock markets of the unon members were closely related to each other compared to non-member countres. Akdogan (1992) employed the asset prcng model to study the proporton of systematc rsks n the asset markets of European Communty (EC). Based on monthly data for , evdence was documented that the convergence of the EC markets was hghly related to the schedule of relaxaton of captal controls n the European Monetary Unon. Usng covarance and correlaton analyss, Johnson and Soenen (1993) and Johnson et al. (1994) also documented smlar conclusons on EC for the same sample perod. Most of the lterature on stock market ntegraton pursued along the drect modelng of the lnkage channels, through contegratng vectors, volatlty spllovers and regme prcng. Soydemr (2000), for example, documented that Mexcan stock market was weakly lnked to those of Argentna and Brazl, but strongly ted wth the US market n the perod of The author concluded that the results could be because these countres are members of dfferent tradng blocs, where Mexco s a member of NAFTA, whle the other two Latn markets spearheaded the Mercado Comun del Cono Sur (MERCOSUR). Chen et al. (2002) also hghlghted that the formaton of MERCOSUR has an mpact on the long-run contegratng relatonshp of the stock prces of Argentna, Brazl, Chle, Colomba, Mexco and Venezuela for the perod In addton, Mexco was found to have domnant nfluence over most of the other Latn markets, due possbly to ts lnkages wth the two domnant NAFTA members, US and Canada. The fndngs are smlar even wth sub-perod analyss anchorng on the Asan and Russan crses, suggestng consstent results across dfferent regmes. These studes are further supported by Johnson and Soenen (2003),

3 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, who used the Geweke measures of feedback. They found evdence of smultaneous lnkages between Canada and Mexco, and also among Argentna, Brazl, Chle (assocate member of MERCOSUR) and Peru usng daly data from 1988 to Although these works offer nsght on the ntegraton of stock markets n countres belongng to a common tradng bloc or a regonal trade arrangement, they are not desgned to examne the tradng bloc effect on asset prcng. Ths paper attempts to nvestgate the effect of trade regonalsm on stock prce determnaton wthn the framework of the nternatonal captal asset prcng model. The focus s on the stock markets of countres that are engaged n three sub-regonal tradng arrangements wthn the Asa-Pacfc Economc Cooperaton (APEC), namely, NAFTA, AFTA and the Australa-New Zealand Closer Economc Relatons Trade Arrangement (CER). 1, 2 As the progress of AFTA, CER and NAFTA can determne the degree of segmentaton wthn APEC, ths study attempts to fnd out f the fnancal ntegraton process has gone beyond regonal tradng agreements by examnng the prcng process of a market n response to the prce dynamcs of the other tradng blocs of whch the gven market s not a member. Ths paper s organzed as follows. The next secton outlnes the asset prcng models and the data employed. Secton 3 presents the results and dscussons on the major fndngs. Concludng comments follow n Secton Methodology The captal asset prcng model (CAPM) offers a theoretcal framework for prcng rsky assets under equlbrum market condton n a domestc context. The nvestors demand for each asset s determned on the bass of perod-by-perod optmzaton of the mean and varance of excess returns (ER): Max U = f E( ER ), V ( ER )], (1) [ t t n order to maxmze ther utlty functon U where ER refers to the returns of the asset over and above the rsk-free rate. In equlbrum, when all nvestors optmal portfolo choces are aggregated, the followng relatonshp s obtaned: E ( Rt mt R ) = β [ E( R R )], or, E cov( Rt, Rmt ) R ) = [ E( Rmt R )], (2) var( R ) ( Rt mt 1 In 1994, APEC declared ts ntenton to form a supraregonal free trade area before Although APEC s stll a trade forum, t s a potental tradng bloc that goes beyond the sze of the European Unon. Ironcally, many blateral free trade agreements contnue to emerge between the members of APEC, and also wth other non-members countres (WTO, 2 Oldest of the three, CER was put n place n ASEAN has agreed on the formaton of AFTA n 1992, whle NAFTA came nto effect n These three regonal groups account for 29 per cent of total ntra-apec trade (APEC, 1997).

4 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, where R s the rate of return to the rsk-free asset, R t s the return to asset and R mt s the total return to the market portfolo. The moments of the expected returns are statonary n Equaton (2) as nvestors are assumed to maxmze the utlty of wealth held for one perod. In realty, multple perods should be consdered and expected returns vary over tme dependng on the nature of the nformaton avalable at a gven pont n tme. The condtonal CAPM s hence more relevant and can be wrtten as: E cov( Rt, Rmt Ω t 1 ) Ω t 1 ) = [ E( Rmt R Ω t )], (3) var( R Ω ) ( Rt R 1 mt t 1 where E(. Ω t 1 ) s the condtonal expectaton based on the nformaton set Ω avalable at tme t-1. Snce non-systematc rsk can be dversfed away, CAPM stpulates that nvestors are rewarded only for the systematc rsk. The nondversfable rsk s measured by beta, whch s determned by the covarance of the return to asset wth the return to the market portfolo. Thus, for expostonal purposes, Equaton (3) thus can be rewrtten as: E ( Rt R t 1 = mt t mt t 1 Ω ) δ cov( R, R Ω ). (4) Gan (2002) provded the theoretcal exposton to show how the CAPM model can be extended to the nternatonal CAPM (ICAPM). In the nternatonal settng, the unverse of the securty portfolo conssts of securtes ssued n dfferent natonal markets. In a fully ntegrated world fnancal market where PPP holds, the expected return of a natonal equty market s solely exposed to the movement n returns of the world portfolo. Extendng the earler arguments on CAPM, the prcng of a natonal equty market s determned by the followng process: E ( Rt R t 1 = Wt t Wt t 1 Ω ) δ cov( R, R Ω ), (5) where R t now refers to the returns n a gven market, and R and R Wt represent the returns on the world rsk-free asset and global portfolo, respectvely, n tme perod t. The expected excess returns of the natonal stock market- are prced on ts covarance wth the world market returns, and on the senstvty coeffcent δ Wt. The ICAPM model mples that the expected excess return of a gven market above the nternatonal rsk-free rate s proportonal to the country specfc but non-dversfable rsk n the world market. A testable form of the ICAPM can be wrtten as follows: R t R = α + β ( R R ) + ε. (6) W Wt t Ths s the tradtonal one-factor verson of the model. Whle α represents the dr W term, β s the beta coeffcent of the th market to the world market and ε t s the market specfc resduals that are orthogonal to the loadng world factor.

5 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, In analyzng the case for EC, Akdogan (1992) replaced the world factor n Equaton (5) wth a weghted average portfolo of the tradng bloc. The model s stll n a onefactor settng gven as follows: E ( Rt R t 1 = Tt t Tt t 1 Ω ) δ cov( R, R Ω ), (7) where R represents the returns to portfolo of assets of all the markets n the bloc. Henceforth, we refer to ths model as the tradng bloc CAPM (TBCAPM). The followng provdes a testable form of the TBCAPM: R t R = α + β ( R R ) + ε. (8) T Tt t Instead of focusng on the tradng bloc alone, we attempt to examne the fnancal ntegraton n a regon gven the exposure to global market movements. To compare the relatve mpact of tradng bloc and world factors, we combne Equatons (6) and (8) nto a two-factor model, that s referred to as the tradng bloc-icapm or TB- ICAPM stated as: R t R = α + β ( R R ) + β ( R R ) + ε. (9) W Wt T Tt t The ncluson of RTt nto ICAPM provdes a loadng factor to capture the effect of the tradng bloc whereby the country s a member. In ths study, Equatons (6), (8) and (9) are estmated for each of the stock markets. In addton, to nvestgate the market response to the dynamcs of the other tradng blocs of whch the gven market s not a member, a mult-factor TB-ICAPM (MTB- ICAPM) s proposed. Extended from Equaton (9), ths model s stated as: R t R = α + β ( R R ) + β ( R R ) + ε, (10) W Wt M M Mt t where M represents all the tradng blocs, ncludng that where the th market s a member. If β M = 0, Equaton (11) collapses to ICAPM. Followng Ramchand and Susmel (1998), all the prcng models are estmated usng the generalzed autoregressve condtonal heteroscedastcty (GARCH) model of Bollerslev (1986). The varance s assumed to be tme varyng, where the error term 2 follows the dstrbuton ε t Ω t 1 ~ N(0, σ t ). The general form of a GARCH(p, q) model s gven by: p σ = ω + φ ε + ψ σ, (11) t j j= 1, t j q m= 1 m, t m where the condtonal varance depends on the squared error terms of p lags (known as the ARCH effects) and the condtonal varance of q lags (known as the GARCH effects). The presence of ARCH and GARCH effects n asset prces are welldocumented n the survey paper by Bollerslev et al. (1992). In ths study, the order of p and q are fxed at one, as ths smple specfcaton s suffcent for most emprcal

6 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, modelng purposes (Engle and Ng, 1993). To account for non-normal condtonal dstrbuton n the resduals, we use the robust quas-maxmum lkelhood estmates suggested by Bollerslev and Wooldrdge (1992). The varance-covarance estmator of ths method s heteroscedastcty consstent. The three tradng blocs of AFTA, CER and NAFTA comprse a total of 10 member countres fve for AFTA (Indonesa, Malaysa, Phlppnes, Sngapore and Thaland), two for CER (Australa and New Zealand), and three for NAFTA (US, Canada and Mexco). As n most of the lterature on ICAPM, monthly data s used. The sample of analyss spans from January 1990 to February El-Hed (2004) ponted out that the condtonal ICAPM should be evaluated over a long perod n order to capture the long-run dynamcs. No sub-perod analyss s conducted n ths study. The country and world stock market ndces compled by Morgan Stanley Captal Internatonal (MSCI) are used to calculate the returns for each country and the global portfolo. The MSCI All Country World Index s a free float-adjusted market captalzaton ndex desgned to measure equty market performance of 49 developed and emergng markets. In computng the return to a tradng bloc portfolo, an equalweghted average of the market returns of all the other member countres except the returns of the market under study s used. Ths s to ensure that the local dynamcs are excluded from the tradng bloc portfolo. The global rsk free rate s proxed by the three-month Treasury bll rates of the US. The Treasury bll rates are downloaded from the webste of the Federal Reserve Bank. 3. Results and Dscusson Table 1 presents the descrptve statstcs and correlaton matrx of all the sample returns seres. It s clear that the uncondtonal dstrbuton of all the seres s not normally dstrbuted, as ndcated by the Jarque-Bera test, except for Australa. The returns of the AFTA members exhbt the most volatle behavor, where Indonesan and Thaland have the hghest standard devaton among all. The returns of the CER and NAFTA members are much less volatle, but Mexco experences hgher volatlty than the others. The results suggest that the large and developed markets have relatvely low volatlty compared to the emergng markets. The relatvely stable characterstcs of the MSCI All Country World Index and 3-month Treasury bll rates are expected. The strongest par-wse correlatons are found for US-Canada and Australa-New Zealand. Among all, Indonesa s the most exogenous market, havng very low correlatons wth non-afta countres. The results also ndcate that the market returns of the same tradng bloc tend to have stronger correlatons compared to the returns of the markets n dfferent tradng blocs. As expected, the US has the hghest correlaton wth the MSCI World Index gven ts most developed and lberalzed stock exchange n the world. Interestngly, all except the return on the US ndex s negatvely correlated wth the 3-month Treasury bll rates. Ths reflects that the other stock markets are adversely affected by an nterest rate rse n the US. The estmated ICAPM, TBCAPM and TB-ICAPM models are reported n Table 2. The estmates for the mult-factor TB-ICAPM model are gven n Table 3. The dagnostc tests n both the tables suggest that all the models are statstcally acceptable. The Jarque-Bera test shows that most of the resdual seres are normally dstrbuted although non-normalty s reported for some cases. The GARCH(1,1)

7 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, specfcaton s generally suffcent to take account of the condtonal heteroscedastcty. 3 The LM test ndcates no further ARCH effect up to twelve lags. At least one of the ARCH or GARCH coeffcents s sgnfcant n the models. The only exceptons are the ICAPM for Phlppnes and TBCAPM for Australa. The GARCH effects are generally larger n magntude compared to the ARCH effects, suggestng that the volatlty s more senstve to the lagged volatlty than to the new surprses n the market. The results from Table 2 suggest that the tradng bloc effects on prcng of a natonal stock market cannot be neglected. The tradng bloc excess returns are statstcally sgnfcant n affectng the excess returns of all the ten ndvdual markets. The exposure to the tradng bloc factor vares by the degree of market openness and development. The magntude of the coeffcent suggests the followng degree of exposure to the AFTA factor: Indonesa, Thaland, Phlppnes, Malaysa and Sngapore. Ths order s consstent n both TBCAPM and TB-ICAPM. The more advanced markets of Sngapore and Malaysa have lower exposure to the AFTA factor. The use of ICAPM alone for the AFTA stock markets can be msleadng. In ths model, Thaland and Indonesa are hghly exposed to the world market movements, and ths s followed by Phlppnes, Sngapore and Malaysa. Ths suggests problems of specfcaton bas n ICAPM due to omsson of the tradng bloc factor. In fact, ncluson of the tradng bloc factor n TB-ICAPM has rendered the world factor nsgnfcant for the smaller markets of Indonesa and Phlppnes. Both Australa and New Zealand exhbt hgh exposure to the world factor n the absence of the CER tradng bloc factor. When CER s factored nto TB-ICAPM, the world beta coeffcent reduced substantally. The stock returns of New Zealand are more exposed to the tradng bloc factor compared to the world factor. On the other hand, the Australa market shows larger exposure to the world factor. For NAFTA, t s clear that the world market s the domnant factor for all the three markets. Ths s consstent across the three asset prcng models. The magntude of the tradng bloc factor n TB-ICAPM s rather small for all three cases and not sgnfcant for Mexco. Gven the economc mportance of NAFTA n terms of world trade, 4 t s not surprsng that the stock markets n NAFTA are hghly exposed to the world factor. To these countres, the world factor and tradng bloc factor are synonymous. Overall, the results show that markets that have a low world beta are lkely to have hgher exposure to the tradng bloc factor, whereas markets wth a hgh world beta tend to have lower tradng bloc exposures. The latter s partcularly true for the developed markets, ncludng Sngapore, Australa and the three markets n NAFTA. Addtonally, we fnd that the tradng bloc factor has sgnfcantly ncreased the explanatory power of the asset prcng models of AFTA. Ths s also true for CER, 3 GARCH(2,1) was estmated for ICAPM of the US and TB-ICAPM of Mexco as the orgnal specfcaton faled to get pass the Q 2 (12) dagnostc test. The second lag of the ARCH coeffcent, however, s not reported n the table. The results are avalable upon request. 4 In 2003, the three countres n NAFTA accounted for 15.9% (19.5%) of total world trade n merchandse (commercal servces) exports, amountng to 1162 (288) bllon USD per annum, and absorbed 22.8% (19.1%) of total world merchandse (commercal servces) mports, amountng to 1727 (229) bllon USD per annum. The fve countres n AFTA only accounted for 5.8% (3.8%) and 5.1% (4.8%) of total world merchandse exports and mports, respectvely. The correspondng fgures for CER are 1.2% (1.5%) and 1.4% (1.5%), respectvely.

8 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, but to a lesser extent. As for NAFTA, the ncrease s only margnal. To fnd the best specfcaton for each market, we rely on the Schwarz nformaton crteron (SC) that mposes a stff penalty on the number of the loadng factors that enter the model. For AFTA, TBCAPM s selected for Indonesa, Malaysa and Phlppnes, whle TB- ICAPM s selected for Sngapore and Thaland. For CER, TB-ICAPM s selected for both Australa and New Zealand. For NAFTA, however, ICAPM cannot be outperformed. We could nfer that the tradng bloc factor s partcularly mportant n the prcng of rsky assets of less developed markets, but the world factor s mportant for the more lberalzed and developed markets. The betas n the multple factor TB-ICAPM reported n Table 3 shed some lght on whether lnkages exst n the prcng of natonal equty among tradng blocs wthn APEC. The frst observaton s that the stock markets n AFTA and CER show strong convergence to ther own tradng bloc factor. Not only the own tradng bloc factor s hghly sgnfcant, the magntude of the coeffcent s also larger than the coeffcent of the world factor and the coeffcent of the other tradng blocs where the gven market s not a member. Consstent wth the earler results, the stock markets n NAFTA converge strongly to the world factor, but less to the own tradng bloc factor. Pockets of cross-bloc relatonshps are sgnfcant. For nstance, the CER factor s sgnfcant for Thaland, the AFTA factor s sgnfcant for New Zealand, the CER factor s sgnfcant for US and Canada, and the AFTA factor s sgnfcant for Mexco. Nevertheless, the magntude of these cross-bloc effects s much smaller compared to the world and the own-bloc effects. Ths ndcates that the tradng blocs have mnmal mpact n nfluencng the market returns of non-member countres. Overall, the result suggests that most of the stock markets converge to ther respectve regonal tradng bloc, and the stock markets of APEC members ncluded n ths study are ndeed segmented by regons.

9 Table 1 Summary of Descrptve Statstcs and Correlaton Matrx of Returns AFTA CER NAFTA World Treasury Bll Indonesa Malaysa Phlppnes Sngapore Thaland Australa New Zealand US Canada Mexco Panel A: Descrptve Statstcs Mean Standard Devaton Skewness Kurtoss Jarque-Bera Probablty Panel B: Correlaton Indonesa Malaysa Phlppnes Sngapore Thaland Australa New Zealand US Canada Mexco World Treasury Bll

10 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, Table 2 Estmates of the Internatonal CAPM (ICAPM), Tradng Bloc CAPM (TBCAPM) and Tradng Bloc-ICAPM (TB-ICAPM) Models Mean Tradng Varance Jarque- ARCH- Adjusted- Log- Schwarz Constant World Bloc Constant ARCH GARCH Q(12) Q 2 (12) Bera LM (12) R 2 Lkelhood Crteron AFTA: Indonesa ICAPM ** ** ** (0.3803) (0.0000) (0.2785) (0.1452) (0.0017) (0.5550) (0.5150) (0.0004) (0.2857) TBCAPM # ** ** ** (0.3305) (0.0000) (0.1560) (0.0073) (0.0000) (0.9320) (0.4350) (0.4551) (0.2640) TB-ICAPM ** ** ** (0.6766) (0.3390) (0.0000) (0.1599) (0.0026) (0.0000) (0.9330) (0.4590) (0.8028) (0.3169) AFTA: Malaysa ICAPM ** ** (0.5413) (0.0000) (0.3481) (0.0522) (0.0000) (0.6840) (0.2540) (0.2752) (0.2698) TBCAPM # * ** * ** (0.0174) (0.0000) (0.3428) (0.0138) (0.0000) (0.5900) (0.4290) (0.6043) (0.5470) TB-ICAPM * ** * ** (0.3112) (0.0217) (0.0000) (0.3183) (0.0169) (0.0000) (0.6410) (0.5440) (0.8265) (0.6443) AFTA: Phlppnes ICAPM ** (0.8091) (0.0000) (0.0605) (0.2421) (0.8320) (0.2240) (0.3310) (0.4775) (0.3497) TBCAPM # ** ** (0.1806) (0.0000) (0.6279) (0.5077) (0.0000) (0.6980) (0.2160) (0.9966) (0.2240) TB-ICAPM ** ** (0.4479) (0.4798) (0.0000) (0.7231) (0.6263) (0.0000) (0.6940) (0.2280) (0.9960) (0.1993) Notes: Fgures n the parentheses are p-values. * and ** denote sgnfcant at the 0.05 and 0.01 levels, respectvely. # Model wth mnmum value of Schwarz crteron. Q(12) and Q 2 (12) refer to the Q-test for sgnfcance of autocorrelaton at lag 12 n the standardzed resduals and squared standardzed resduals, respectvely.

11 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, Table 2 (Contnued) Estmates of the Internatonal CAPM (ICAPM), Tradng Bloc CAPM (TBCAPM) and Tradng Bloc-ICAPM (TB-ICAPM) Models Mean Tradng Varance Jarque- ARCH- Adjusted- Log- Schwarz Constant World Bloc Constant ARCH GARCH Q(12) Q 2 (12) Bera LM (12) R 2 Lkelhood Crteron AFTA: Sngapore ICAPM ** * ** (0.9249) (0.0000) (0.2011) (0.0410) (0.0000) (0.6030) (0.1140) (0.1243) (0.2293) TBCAPM ** ** ** ** (0.0002) (0.0000) (0.3711) (0.0575) (0.0000) (0.5940) (0.2910) (0.0003) (0.4225) TB-ICAPM # ** ** * ** * (0.5581) (0.0000) (0.0000) (0.2736) (0.0359) (0.0000) (0.8270) (0.2340) (0.0370) (0.2281) AFTA: Thaland ICAPM ** ** * (0.1135) (0.0000) (0.3866) (0.0714) (0.0000) (0.0330) (0.1990) (0.6783) (0.4322) TBCAPM ** ** ** (0.4165) (0.0000) (0.4245) (0.0802) (0.0000) (0.6720) (0.6820) (0.0000) (0.7916) TB-ICAPM # * ** ** ** ** (0.0422) (0.0039) (0.0000) (0.4052) (0.0664) (0.0000) (0.6870) (0.4080) (0.0000) (0.5449) Notes: Fgures n the parentheses are p-values. * and ** denote sgnfcant at the 0.05 and 0.01 levels, respectvely. # Model wth mnmum value of Schwarz crteron. Q(12) and Q 2 (12) refer to the Q-test for sgnfcance of autocorrelaton at lag 12 n the standardzed resduals and squared standardzed resduals, respectvely.

12 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, Table 3 Estmates of the Mult-factor Tradng Bloc ICAPM models Mean Varance Jarque- ARCH- Adjusted- Log- Schwarz Constant World NAFTA CER AFTA Constant ARCH GARCH Q(12) Q 2 (12) Bera LM (12) R 2 Lkelhood Crteron AFTA Indonesa * ** ** ** (0.7132) (0.0394) (0.1027) (0.5414) (0.0000) (0.1562) (0.0035) (0.0000) (0.9130) (0.4370) (0.7481) (0.3052) Malaysa ** * ** (0.2901) (0.0701) (0.8130) (0.3755) (0.0000) (0.3351) (0.0124) (0.0000) (0.5450) (0.5730) (0.7490) (0.6202) Phlppnes ** ** (0.4487) (0.6213) (0.5111) (0.2426) (0.0000) (0.7062) (0.6353) (0.0000) (0.6500) (0.2670) (0.9999) (0.2669) Sngapore ** ** * ** (0.5186) (0.0000) (0.2114) (0.0931) (0.0000) (0.2651) (0.0195) (0.0000) (0.6800) (0.1110) (0.4173) (0.1145) Thaland ** * ** * ** ** (0.0350) (0.2165) (0.7011) (0.0373) (0.0000) (0.4273) (0.0465) (0.0000) (0.7300) (0.2930) (0.0000) (0.4138) Notes: Fgures n the parentheses are p-values. * and ** denote sgnfcance at the 0.05 and 0.01 levels, respectvely. Q(12) and Q 2 (12) refer to the Q-test for sgnfcance of autocorrelaton at lag 12 n the standardzed resduals and squared standardzed resduals, respectvely.

13 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, Table 3 (Contnued) Estmates of the Mult-factor Tradng Bloc ICAPM models Mean Varance Jarque- ARCH- Adjusted- Log- Schwarz Constant World NAFTA CER AFTA Constant ARCH GARCH Q(12) Q 2 (12) Bera LM (12) R 2 Lkelhood Crteron CER Australa ** ** * ** (0.6909) (0.0008) (0.0586) (0.0000) (0.3178) (0.2170) (0.0264) (0.0000) (0.1150) (0.8770) (0.5945) (0.6032) New Zealand ** ** ** ** ** (0.6937) (0.0081) (0.2691) (0.0000) (0.0085) (0.3954) (0.8234) (0.0000) (0.0050) (0.5090) (0.6723) (0.5490) NAFTA US ** ** ** ** ** * (0.2617) (0.0000) (0.0008) (0.0000) (0.6959) (0.0000) (0.0009) (0.0308) (0.4460) (0.4310) (0.3118) (0.5756) Canada ** * ** ** (0.4752) (0.0000) (0.0223) (0.0030) (0.0903) (0.3156) (0.0784) (0.0005) (0.5460) (0.9320) (0.3295) (0.9721) Mexco ** ** ** ** ** (0.0006) (0.0066) (0.3218) (0.3624) (0.0005) (0.2447) (0.0549) (0.0000) (0.8150) (0.9830) (0.0000) (0.9878) Notes: Fgures n the parentheses are p-values. * and ** denote sgnfcance at the 0.05 and 0.01 levels, respectvely. Q(12) and Q 2 (12) refer to the Q-test for sgnfcance of autocorrelaton at lag 12 n the standardzed resduals and squared standardzed resduals, respectvely.

14 4. Concluson Ths paper attempts to nvestgate the mpact of regonalsm on asset prcng of natonal stock markets. The prcng convergence of stock markets n the three subregonal tradng blocs of AFTA, CER and NAFTA wthn APEC s examned usng ICAPM that allows for tme varyng volatlty. The results hghlght the mportance of the tradng bloc effect n asset prcng, especally for small and emergng markets. Asset prcng models that ncorporate only the world factor are more relevant for the large and developed markets. The nternatonal CAPM s suffcent for the markets n NAFTA, but the modfed CAPM wth the tradng bloc factor generally fts better and provdes hgher explanatory power for AFTA and CER. The exposure to the tradng bloc factor s very hgh for the markets n AFTA and CER, but the mpact of the world factor s equally promnent n the relatvely developed markets such as Sngapore and Australa. The mpact of trade regonalsm should not be neglected n nternatonal asset prcng models to avod specfcaton problem. The mpact of cross-bloc effects on asset prcng s rather mnmal. In other words, the prcng of a stock market s hghly nfluenced by the own-bloc and/or world factor, but prcng dynamcs of the other tradng blocs of whch ths market s not a member have very lttle nfluence. The results suggest that the regonalsm s a possble explanaton for the cause of segmentaton among the fnancal markets wthn the APEC regon. Ths study s lmted to a selected number of trade agreements wthn APEC. A smlar study can be conducted for the other tradng blocs wthn and outsde APEC. The models ncluded n the study do not take nto account of dosyncratc rsks of ndvdual markets. The mldly segmented ICAPM framework can be consdered for future research to overcome ths shortcomng. A detaled senstvty analyss of the beta coeffcents, whch are not ncluded n ths study s also useful for understandng the stablty of the underlyng parameters. Acknowledgements An earler verson of ths paper s presented at the Internatonal Conference n Economcs and Fnance, May 2005, Labuan, Malaysa. We thank the conference partcpants, n partcular, Tan Hu Boon, Mansor Ibrahm and Kok Km Lan for valuable comments. We are also grateful to an anonymous referee for helpful comments and suggestons. References Akdogan, H. (1992) Behavour of systematc rsk n a regonally ntegrated model for stock prces. Economcs Letters, 39: Asa-Pacfc Economc Cooperaton. (1997) The Impact of Subregonalsm n APEC, Sngapore: APEC. Bollerslev, T. (1986) Generalzed autoregressve condtonal heteroskedastcty. Journal of Econometrcs, 31:

15 Hooy and Goh / Labuan Bulletn of Internatonal Busness & Fnance, 3, 2005, Bollerslev, T. and Wooldrdge, J.M. (1992) Quas-maxmum lkelhood estmaton and nference n dynamc models wth tme varyng covarances. Econometrc Revews, 11: Bollerslev, T., Chou, R.Y. and Kroner, K.F. (1992) ARCH modelng n fnance: A revew of the theory and emprcal evdence. Journal of Econometrcs, 52: Chan, K.C., Karoly, G.A., Stulz, R.M. Global fnancal markets and the rsk premum on US equty. Journal of Fnancal Economcs, 32: Chen, G.M., Frth, M. and Ru, O.M. (2002) Stock market lnkages: evdence from Latn Amerca. Journal of Bankng and Fnance, 26: Engle, R.F. and Ng, V.K. (1993) Measurng and testng the mpact of news on volatlty. Journal of Fnance, 48: Fratzscher, M. (2002) Fnancal market ntegraton n Europe: On the effects of EMU on stock markets. Internatonal Journal of Fnance and Economcs, 7: Gan, W.B. (2002) Macroeconomc factors and the prcng of rsk n East Asan equty markets. Malaysan Journal of Economc Studes, 39: Johnson, R. and Soenen, L. (1993) Stock Market Reacton to EC Economc and Monetary Intenton. European Management Journal, 11(1): Johnson, R. and Soenen, L. (2003) Economc ntegraton and stock market comovement n the Amercas. Journal of Multnatonal Fnancal Management, 13: Johnson, R., Lndvall, J. and Soenen, L. (1994) EC economc and monetary ntegraton: Implcatons for European equty nvestors. European Management Journal, 12(1): Lessard, D.R. (1973) Internatonal portfolo dversfcaton: A Multvarate analyss for a group of Latn Amercan countres. Journal of Fnance, 28: Ramchannd, L. and Susmel, R. (1998a) Varances and covarances of nternatonal stock returns: The nternatonal captal asset prcng model revsted. Journal of Internatonal Fnancal Markets, Insttutons and Money, 8: Soydemr, G. (2000) Internatonal transmsson mechansm of stock market movements: Evdence from emergng equty markets. Journal of Forecastng, 19: