Assignment Help!!!

International Economic Journal Vol. 22, No. 1, March 2008, 113–126

Does trade regionalism increase stock market segmentation within

a trading bloc? Chee Wooi Hooy*

Finance Section, School of Management, University Sains Malaysia, Penang, Malaysia

(Received 29 June 2006; final version received 7 February 2007 )

Whether trade regionalism remains a stumbling block to economic globalization remains a debatable issue. This paper investigates the impact of trade regionalism on stock market segmentation for the case of AFTA, EU and NAFTA. We exploit the pricing error of the trading-bloc capital asset pricing model (TB-CAPM) as an indicator for stock market seg- mentation based on Akdogan (1992) and Adler and Qi (2003). We use intra-trade ratio as the indicator for trade regionalism. Controlling for other determinants for market integration, evidence is found that trade regionalism helps to explain stock market segmentation in EU and NAFTA, which are both dominated by developed countries. For NAFTA, the higher the intra-trade ratio the lower the degree of market segmentation within the trading bloc. For EU however, the intra-trade ratio contributes to higher stock market segmentation.

Keywords: stock market segmentation; trade regionalism; TB-CAPM; intra-trade ratio

Introduction

The global trend in trade regionalism has entered a new era. The number of trading blocs formally registered with WTO has been accelerating over the last few years. Before the 1990s, only 27 agreements were established. The number jumped to a surprising 162 by 2005, and continues to surge to a total of 205 as of July 2007.1 A hot debate on this phenomenon of increasing regionalism is whether it will hinder or promote a mul- tilateral trading system. Many scholars, including Bhagwati (1993) and Frankel et al. (1995) found that the recent trade regionalism is more likely to be welfare reducing and works against world economic globalization. WTO, however, believes that such development is a complementary progress as long as it can help trade flow more freely

*Email: [email protected]

ISSN 1016-8737 print/ISSN 1743-517X online © 2008 Korea International Economic Association DOI: 10.1080/10168730801887083 http://www.informaworld.com

114 C. W. Hooy

within the bloc, without barriers being raised on trade outside the bloc (see GATT Article 24). Regionalism is seen as an instrument that allows groups of countries to negotiate rules and commitments that go beyond what are possible multilaterally, in the hope that they will develop into topics of discussion in WTO later. This view is attributed mainly to MacMillan (1993).

In principle, when real sectors get more integrated, capital markets will become less segmented. The intuitive sense is that international finance serves to facilitate international trade. When a country becomes more integrated with their trading bloc counterparts in terms of trade, it will take an initiative to lock-in economic and political reforms to accelerate the transition to a market economy (Hoekman and Kostecki, 2001). This will further lead to policy coordination in both monetary and fiscal aspects among the member countries. With increasing real sector integration, intra-bloc capital flows are likely to be enhanced across borders. The liberalization effort will also reduce cross-market frictions, and the law of one price will play its role to prevent investors from taking advantage of any pricing discrepancies among the bloc members. If the WTO is correct that trade regionalism will promote globalization, then the integration process goes beyond the bloc basis. This means that the reduction in capital market segmentation will happen not only within the bloc, but at the same time, with the rest of the world.

Unfortunately, empirical works relating trade regionalism to stock market integration/segmentation remain scarce in the literature. This is probably due to the fact that trade flow information is often not incorporated in the investor’s pric- ing decision. The impact of trade flow can always prevail through various economic fundamentals because they are more easily accessed by the investors. Therefore, macroeconomics-oriented variables are often taken as quite reasonable factors in explaining stock market integration. Among the macroeconomic variables that have been considered in asset pricing models include trade openness, business cycle, exchange rates and interest rates. In addition, others also explore the role of microstruc- ture oriented variables such as dividend yield, liquidity, book-to-market, or even market volatility to explain stock market integration.2

The works of Heaney, Hooper and Jaugietis (2000) and Heaney and Hooper (2001) are the only studies we found to relate integration of the stock market with trading bloc grouping. Both papers utilized cluster analysis to examine whether stock markets form a grouping in a hierarchical dendogram and they recorded clear evidence of world stock market segmentation that is very much consistent with the existence of regional trading bloc and economic ties between the sample countries. Many other market integration studies have pointed out the role of trade regionalism on market integration. Nevertheless, it was not in their objective to test for the direct relationship between the two. These include Heaney and Hooper (1999) and Ng (2002) on AFTA; Akdogan (1992), Corhay, Rad and Urbain (1993), Johnson and Soenen (1993), Johnson, Lindvall and Soenen (1994), Monadjemi and Perry (1996), Choudhry (1996), Kanas (1998) and Fratzscher (2002) on EMU; Soydemir (2000), Seabra (2001), Edwards and Susmel (2001), Chen, Firth and Rui (2002), Heaney, Hooper and Jaugietis (2002) and Johnson and Soenen (2003) on MERCOSUR; and Adler (1995), Ewing, Payne and Sowell (1999), Adler and Qi (2003) on NAFTA. These works have provided strong grounds for further empirical investigation on this issue, but call for a more direct approach.

The focus of this paper is to explore how trade regionalism affects the segmentation of stock markets within the trading bloc. We aim to find out if trade regionalism has helped to integrate stock markets among the trading members, or whether it has induced more segmentation among the stock markets within the trading bloc.

International Economic Journal 115

The empirical findings of this paper allow the macroeconomic policymakers who are engaged in the trading bloc agreements to understand better the likely impact of real sector convergence on the financial sector integration. In addition, the findings allow the related central banks and security exchanges to take note of the sources of their asset pricing divergences with those of the trading-bloc binding member countries.

This paper contributes in two ways. First, we use a mix of a determinant model for market integration, and a cross-section asset pricing approach to investigate the impact of trade regionalism on trading bloc stock market segmentation. Our approach is different from the cluster analysis of Heaney, Hooper and Jaugietis (2000) and Heaney and Hooper (2001). Our model has controlled other important determinants in the market integration process. This leads to the second contribution of this paper. The modeling of trade regionalism as a determinant for market segmentation/integration has extended the determinant model used in Carrieri, Errunza and Hogan (2007) in explaining stock market integration. In fact, we find that trade regionalism plays a relatively important and significant role in explaining stock market segmentation within a trading bloc, as compared to the other integration determinants suggested in Carrieri, Errunza and Hogan (2007).

This paper is organized as follows. The next section outlines the framework of analysis and the data employed. The third section presents the statistical results and discussions on the major findings. Concluding comments are in the final section of the paper.

Methodology and data

Empirical indicators are used to measure the degree of trade regionalism and stock market segmentation in a trading bloc. The level of trade regionalism is represented by the intra-trade ratio, which is referred to as the Trade Regionalism Index (TRI). The TRI for trading bloc T is defined as:

TRIT,t = 1

nT

nT∑

i=1

BTTit WTTit

(1)

where nT is the number of member countries in trading bloc T, BTTit is the total bilateral import and export of country-i with the other member countries of the same bloc, and WTTit is the total trade (import plus export) of country-i with the world market. This base measure for goods movement serves to compare real integration within a trading bloc in comparison to total movement to the world market.3

7 According to Korajczyk (1996), pricing errors estimated within the framework of some form of asset pricing model can be used to measure how segmented a subject market is to the world market. Korajczyk (1996) developed an index to capture the dynamic integration states of market integration from the pricing errors estimated under the framework of International Arbitrage Pricing Theory (IAPT). The idea was applied and extended in Levine and Zervos (1998) using both IAPT and World Capital Asset Pricing Model (WCAPM) to study stock market development and economic growth.

In this paper, we are interested in measuring stock market segmentation among members that bind under a trade agreement. A variant of asset pricing model applied by Akdogan (1992) and Adler and Qi (2003) is appropriate for this purpose. The model,

116 C. W. Hooy

which we shall term as Trading-Bloc Asset Pricing Model (TB-CAPM henceforth) asserts that the pricing of cost of capital can be determined in a linear return-generating process given as follows:

rit = αi + βirTB,t + εit ; ∀i (2) where rit represent the excess market returns of individual country-i and rTB,t represent the excess returns of a portfolio of equal weighted stock indices from a trading bloc. Given that investors are mean-variance optimizers, the beta βi measures the systematic risk of stock market-i with respect to the trading-bloc portfolio movements. The error term εit captures the idiosyncratic risk that is orthogonal to the trading bloc portfolio. The aggregate relative risk aversion of individual investors is assumed to be constant (Chan, Gup and Pan 1992). Akdogan (1995) and Adler and Qi (2003) have explored the use the above TB-CAPM setting in pricing the ‘regionalism effect’ in EU and NAFTA, respectively.

With the assumption of perfect trading bloc integration, the intercept αi in model (2) should be equal to zero. This is essentially another version of Black’s (1972) zero-beta CAPM. If αi is significantly different from zero, the hypothesis of perfect integration with the trading bloc is rejected. The pricing error captures the deviation from an integration state. It is therefore appropriate to interpret it as a measure of the degree of segmentation among the stock markets in the trading-bloc portfolio.

The pricing error is used to construct an indicator for trading-bloc stock market segmentation. Following Levine and Zervos (1998), the pricing error is adjusted using the following transformation:

SMSIT = − |αT| (3) where αT =

∑nT i=1(αi/nT). We refer to this as the stock market segmentation index

(SMSI). By making the adjustment, the SMSI is negatively correlated with the degree of segmentation. The value of the index decreases with higher official barriers and taxes to international asset trading, bigger transaction costs, and larger impediments to the flow of firm information (see Korajczyk, 1996). Therefore, the measure indicates that a higher value of SMSI corresponds to a lower degree of market segmentation. Perfect integration with the world market occurs when SMSI is equal to zero.

There are various econometric techniques that can be used to generate a stochastic time series for SMSI. While Korajczyk (1996) used a simple rolling approach, Levine and Zervos (1998) only estimated a static model for cross-sectional pricing errors. In this paper, we follow a popular cross-sectional regression method suggested by Fama and MacBeth (1973). This is a two-step procedure based on rolling time-series regression. In the first step, we estimate model (2) for t = 1, 2, . . . , N, where N is the number of observations for the first 5 years, and i = 1, 2, . . . , nT. The estimated beta, β̂TBi , is obtained, and treated as the ex-ante beta for period N + 1, denoted as β̂Wi,N+1. In the second step, the ex-post excess return for period N + 1 is regressed on the ex-ante beta. The following cross-sectional regression is estimated for i = 1, 2, . . . , nT:

ri,N+1 = αN+1 + γN+1β̂TBi,N+1 + εi,N+1 (4)

From here, the pricing error is obtained and SMSIT,N+1 is computed. The two steps are repeated by moving one observation down the 5-year window, i.e. for t = 2, 3, . . . , N + 1 to compute SMSIT,N+2. The time series SMSIT,t is generated by repeating the whole

International Economic Journal 117

process until the last observation. After computing SMSIT,t and TRIT,t for each trading bloc, we estimate the following model:

SMSI = a + b1TRI + b′Z + ε (5) In model 5, the SMSI is regressed to the TRI, controlling for a set of information related to the stock market integration process, as captured by the vector of Z.4 Following Carrieri, Errunza and Hogan (2007), we included three variables in the information set:

Z = f (TO, FD, MV) where TO, FD and MV are trade openness, financial development (proxies by market capitalization to GDP) and market volatility, respectively.

In the exploration work to find the possible determinants for stock market inte- gration, Carrieri, Errunza and Hogan (2007) uses TO (trade to GDP) to proxy for macroeconomic development or economic integration (with the world); FD (stock market capitalization to GDP) to proxy for the extent of financial market develop- ment; two additional measures to capture financial liberalization; and world volatility as a control variable. The two measures for financial liberalization adopted by them are the dummy set borrowed from Bekaert and Harvey (2000) and Kaminsky and Schmuk- ler (2001). However, the liberalization dates provided by Bekaert and Harvey (2000) are limited to 20 emerging markets, while the dummy set provided by Kaminsky and Schmukler (2001) only covered until 2002.5 For the above reason, we only introduced TO, FD and MV into the information set, and excluded the financial liberalization variables. In addition, the MV that we used is the average volatility of the trading bloc portfolio proxy by the conditional volatility series generated using a GARCH (1, 1) process.6 Our final market segmentation model takes the following form:

SMSI = a + b1TRI + b2TO + b3FD + b4MV + ε (6) We expect the slope coefficient b1 to be negatively significant if trade regionalism signif- icantly increases stock market segmentation among the trading bloc members. On the other hand, if b1 is positively significant, this indicates that trading-bloc economic inte- gration contributes to reduce stock market segmentation among the members. Using a cross-sectional time-series approach, Carrieri, Errunza and Hogan (2007) found that TO poses significant negative effects on the world integration of eight emerging market, while FD has significant positive effects. The loading for world volatility is inconclusive. In our case, as we are studying three trading blocs averaged from a mix of both devel- oped and emerging countries trading blocs, the above result at best serves as a guide. We estimate model (6) using Newey–West heteroskedasticity and autocorrelation consistent covariance estimates for evaluating the significance of the coefficients.

We examine three notable trading blocs – the European Union or EU (consisting of 14 markets), North American Free Trade Area or NAFTA (consisting of three mar- kets) and ASEAN Free Trade Area or AFTA (consisting of five markets). As shown in Table 1, these three trading bloc are at a different level of economic integration. EU is a monetary union; NAFTA is a free trade area, while AFTA is established on the basis of a preferential trade agreement. The level of trade regionalism for each case is clear.

The liberalization dates for the list of stock markets under each trading bloc are tabu- lated in Table 2. The dates are based on two popular papers, Kaminsky and Schmukler (2003) and Bakeart, Harvey and Lundblad (2005), which complement each other to provide us with a complete picture of the stock market liberalization process in the

118 C. W. Hooy

Table 1. Summary of trading bloc information.

GATT/WTO notification

Trading Bloc Date of entry Type of Agreement into force Date Related provisions agreement

EC (Treaty of Rome) 1-Jan-58 10-Nov-95 GATS Art. V Services NAFTA 1-Apr-94 1-Mar-95 GATS Art. V Services NAFTA 1-Jan-94 1-Feb-93 GATT Art. XXIV Free trade AFTA 28-Jan-92 30-Oct-92 Enabling Clause Preferential arrangement

Source: http://www.wto.org/.

Table 2. Stock market liberalization date.

Trading Bloc Countries Kaminsky–Schmukler (2003) Bakeart et al. (2005)#

EU Austria – ∗ Belgium – ∗ Denmark Pre 73 ∗ Finland Pre 73p/Jan 90 ∗ France Pre 73 ∗ Germany Pre 73 ∗ Greece – 1987 Ireland Pre 73p/Jan 92 ∗ Italy Pre 73 ∗ Netherlands – ∗ Portugal Pre 73-Dec 75/Jan 86 1986 Spain Pre 73 1985 Sweden Pre 73p/Jan 80 ∗ UK Pre 73 ∗

NAFTA Canada Pre 73 ∗ Mexico Jan 89p/Jan 91 1989 USA Pre 73 ∗

AFTA Indonesia Dec 88p/Aug 89 1989 Malaysia Jul 73/Jan 75p/Dec 84-Dec 97 1988 Philippines Mar 86p/Jan 94 1991 Singapore – ∗ Thailand Jan 88p/Jan 90 1987

Note (from the sources): #the official liberalization dates, date of the first American Depository Receipt (ADR) issuance, and first country fund are based on Bekaert and Harvey (2000); Pre 73 (Pre 73p) means that the sector is already fully (partially) liberalized at that time, with no significant measures taken at that date; ∗fully liberalized (industrialized). Source: (1) Kaminsky and Schmukler (2003, Table 1); (2) Bakeart et al. (2005, Appendix A).

three trading blocs under our study. Basically, almost all of the EU markets have been liberalized before the breakdown of the Bretton Wood exchange rate regime in 1973. For NAFTA, both Canada and the US markets are highly liberalized long before NAFTA came into force in mid-1994. However, Mexico only started to liberalize her stock mar- kets in 1989 (partially). In the case of AFTA, except Malaysia and Singapore, the other three AFTA members begin their liberalization process in the late 1980s. This implies that the integration of NAFTA and AFTA stock markets might still in an untimely stage be compared to EU.

The monthly stock returns are computed from the stock prices sourced from Morgan Stanley Capital International (MSCI) for the period January 1988 to May 2005. The world portfolio is represented by the MSCI All Country World Index. The US 3-month

International Economic Journal 119

Treasury bill rates are used to represent the risk free rates in the computation of excess returns in the TB-CAPM model. The trade data are extracted from the IMF Direction of Trade Statistics, while market capitalization, nominal GDP and the US 3-month Treasury bill rates are all taken from DataStream database in the Watson Library of Columbia University in New York.

Results and discussion

The summary of descriptive statistics for all the variables is reported in Table 3. The SMSI for AFTA has the lowest negative mean and it is highly volatile, while the SMSI for EU and NAFTA are of similar size and variation. This reflects that the stock markets of AFTA, which is dominated by emerging markets, are more segmented from each other relative to those in the EU and NAFTA, which are dominated by developed markets. In terms of economic integration, both EU and NAFTA are highly integrated relative to AFTA. TRI values of EU and NAFTA indicate that intra-trade among the members account for 60.73% and 62.03% of their total trade, respectively, while TRI of AFTA is only 18.86%. However, in terms of economic openness, AFTA’s trade to GDP ratio is relatively high (43.06%) compared with the two developed trading blocs (21.30% and 21.29% for EU and NAFTA, respectively), but the openness measure for AFTA is also highly unstable, as indicated by its range and standard deviation. In terms of financial development (market capitalization to GDP), NAFTA is far ahead of both the EU and AFTA. The average financial development in AFTA is slightly higher than that in the EU, but the financial development in the EU is the most stable among the three. The highly unstable FD in AFTA is due possibly to the fast emergences of their stock markets (the four Asian tigers) during the 1990s, but the AFTA members have suffered a huge drawback during the Asian financial crisis in the late 1990s. This is reflected in the market volatility measure where AFTA is found to have a relatively high value of the mean and standard deviation, relative to EU and NAFTA. Finally, all these indices are not normally distributed, except the TRI of EU and NAFTA.

Table 3. Descriptive statistics.

Mean Maximum Minimum Std Dev Skewness Kurtosis Jarque-Bera

SMSIAFTA −0.1009 −0.0012 −0.5584 0.0909 −1.9365 8.3661 271.8957 (0.0000)∗∗∗ SMSIEU −0.0643 −0.0001 −0.2498 0.0508 −1.0462 3.7713 30.8719 (0.0000)∗∗∗ SMSINAFTA −0.0629 −0.0002 −0.2929 0.0496 −1.3504 5.5372 85.2514 (0.0000)∗∗∗ TRIAFTA 0.1886 0.2188 0.1548 0.0143 −0.3301 2.4098 4.8676 (0.0877)∗ TRIEU 0.6073 0.6439 0.5530 0.0175 −0.2871 2.8983 2.1117 (0.3479) TRINAFTA 0.6203 0.6515 0.5655 0.0187 −0.3471 2.5821 4.0762 (0.1303) TOAFTA 0.4306 0.6564 0.2609 0.0979 0.6103 2.3219 12.1058 (0.0024)

∗∗∗ TOEU 0.2130 0.2988 0.1376 0.0379 0.1892 2.1980 4.8822 (0.0871)∗ TONAFTA 0.2129 0.2859 0.1660 0.0250 0.5626 3.2219 8.1653 (0.0169)

∗∗ FDAFTA 2.7971 4.9881 1.0218 1.1115 0.6283 1.8917 17.4300 (0.0002)

∗∗∗ FDEU 2.1902 3.4018 0.9900 0.6310 −0.1266 1.9626 7.0792 (0.0290)∗∗ FDNAFTA 4.7212 8.3163 2.3626 1.4537 0.2584 2.1956 5.6761 (0.0585)

∗ VOLAFTA 3.7233 13.5686 1.0096 2.1497 1.6170 6.1413 126.1921 (0.0000)

∗∗∗ VOLEU 2.3016 6.0847 1.0072 0.9946 1.4469 5.0433 77.9090 (0.0000)∗∗∗ VOLNAFTA 2.3162 9.1779 0.6152 1.2368 2.0954 10.8738 493.9363 (0.0000)

∗∗∗

Figures in the parentheses are probability values. Std Dev denotes standard deviation.

120 C. W. Hooy

Table 4. Correlation coefficients.

SMSIAFTA SMSIEU SMSINAFTA TRIAFTA TRIEU TRINAFTA

SMSIAFTA 1 SMSIEU 0.0780 1 SMSINAFTA −0.0137 0.0835 1 TRIAFTA 0.2245 0.0596 −0.1064 1 TRIEU 0.1655 −0.1824 0.0610 −0.2826 1 TRINAFTA −0.1198 −0.2979 0.1832 0.0053 0.0066 1

Due to space limitations, we have only tabulated the pairwise correlation coefficients of SMSI and TRI indices in Table 4. They are the two target subjects of our study. Gener- ally, the correlations are rather weak among all the indices. The strongest positive and negative pairs are SMSIAFTA– TRIAFTA (0.2245) and SMSIEU–TRINAFTA(−0.2979), respectively.

Both the SMSI and TRI indices are to measure the integration of real and financial sectors, respectively. Owing to increasing globalization and liberalization of the world economy over the last few decades, these two variables might move in the same direction with an upward trend. A similar problem might happen as well for trade openness, financial development and market volatility. If this is the case, a simple regression of these variables would capture only a spurious relationship among them. Furthermore, the 1997 Asian financial crisis, which developed into world-scale turmoil in 1998–1999 might induce a structural break in the time series. To account for this setback, we run a preliminary analysis on all the variables using a constant, time trend and binary dummy variable that takes the value of one for the period July 1997–December 1999, and zero otherwise, as shown in model (7):

Y = δ0 + δ1Trend + δ2D97 + e (7) where Y denotes all the time series involved: SMSI, TRI, TO, FD and MV. δ0, δ1 and δ2 are the parameters to be estimated.

The results of model (7) on all the time series are reported in Table 5. Generally, the time series are highly loaded to the time trend and 1997 dummy (or either one, except for TOEU and MVEU). What this suggests is that they are possibly non-stationary behaved time series. As a result, we decided to replace the original series with the residual series collected from model (7). These residual series represent the original time series after removing the mean, time trend, and the 1997 structural break, allowing us to avoid the spurious regression problem in OLS. To be cautious, we ran two unit-root tests on these new series and the result is reported in Table 6. Both the Augmented Dickey–Fuller and Phillips–Perron unit root tests support that the new series are stationary in level.7

The estimated results for model (6) are shown in panel A of Table 7.8 The coefficient estimate of TRI (b1) is positive for AFTA and NAFTA, but we only have statistical evidence (at 5%) to believe that the coefficient b1 of NAFTA is different from zero. For EU, b1 is negative and statistically significant. The result implies that the effect of trade regionalism is only valid in the developed-countries dominated trading blocs but the impact is different for the American and the European blocs.

The significant positive b1 for NAFTA highlights that as the economic integration has become greater, there has been less segmentation of the stock markets among Canada, Mexico and the US. The pricing of equity indices to the trading bloc generally get more synchronized. The significant negative b1 for EU indicates otherwise. Gen- erally, as EU members get higher trade exposure to each other, the statistical evidence

International Economic Journal 121

Table 5. Regressions for trend and structural break.

Model: Y = δ0 + δ1Trend + δ2D97 + e (7) where Y refers to all the variables understudy: the Stock Market Segmentation Index (SMSI), Trade Regionalism Index (TRI), Trade Openness (TO), Financial Development (FD) and Market Volatility (MV). Trend is a time trend variable while D97 is a binary dummy takes value of 1 for July 1997 – Dec 1999, and zero otherwise.

Variables Trading Bloc Constant Trend D97 Adj-R2

SMSI AFTA −0.1251∗∗∗ (0.0149) 0.0004∗∗∗ (0.0002) −0.0447∗∗ (0.0179) 0.0776 EU −0.0647∗∗∗ (0.0086) 0.0001 (0.0001) −0.0181∗ (0.0104) 0.0102

NAFTA −0.0552∗∗∗ (0.0083) −0.0002 (0.0001) 0.0178∗ (0.0100) 0.0275 TRI AFTA 0.1664∗∗∗ (0.0010) 0.0003∗∗∗ (0.0000) −0.0005 (0.0012) 0.8212

EU 0.6178∗∗∗ (0.0028) −0.0001∗∗∗ (0.0000) −0.0100∗∗∗ (0.0034) 0.1124 NAFTA 0.6186∗∗∗ (0.0027) 0.0000 (0.0000) 0.0242∗∗∗ (0.0033) 0.2759

TO AFTA 0.5412∗∗∗ (0.0130) −0.0013∗∗∗ (0.0001) −0.0794∗∗∗ (0.0156) 0.3923 EU 0.2188∗∗∗ (0.0065) −0.0001 (0.0001) 0.0034 (0.0078) −0.0020

NAFTA 0.2386∗∗∗ (0.0035) −0.0003∗∗∗ (0.0000) −0.0124∗∗∗ (0.0043) 0.3090 FD AFTA 4.2839∗∗∗ (0.1268) −0.0172∗∗∗ (0.0014) −1.0446∗∗∗ (0.1521) 0.5530

EU 1.3247∗∗∗ (0.0643) 0.0095∗∗∗ (0.0007) 0.8168∗∗∗ (0.0772) 0.6429 NAFTA 2.8189∗∗∗ (0.1731) 0.0228∗∗∗ (0.0019) 1.0561∗∗∗ (0.2077) 0.5129

MV AFTA 3.1839∗∗∗ (0.2749) −0.0023 (0.0031) 3.5425∗∗∗ (0.3298) 0.4383 EU 2.0468∗∗∗ (0.1684) 0.0025 (0.0019) 0.3328 (0.2021) 0.0150

NAFTA 2.2572∗∗∗ (0.2026) −0.0016 (0.0023) 0.8951∗∗∗ (0.2432) 0.0777 Note: Figures in the parenthesis are the standard error. Adj-R2 represent adjusted R2 .

Table 6. Unit root tests (without trend and intercept).

Variables Trading Bloc Augmented Dickey–Fuller Phillips–Perron

SMSI AFTA −13.0394 (0.0000)∗∗∗ −13.0324 (0.0000)∗∗∗ EU −12.4314 (0.0000)∗∗∗ −12.5947 (0.0000)∗∗∗

NAFTA −10.1668 (0.0000)∗∗∗ −10.1556 (0.0000)∗∗∗ TRI AFTA −8.0366 (0.0000)∗∗∗ −8.3633 (0.0000)∗∗∗

EU −5.1001 (0.0000)∗∗∗ −5.2534 (0.0000)∗∗∗ NAFTA −1.3007 (0.1779) −5.3733 (0.0000)∗∗∗

TO AFTA −2.4763 (0.0133)∗∗ −3.7594 (0.0002)∗∗∗ EU −2.0938 (0.0353)∗∗ −3.0598 (0.0024)∗∗∗

NAFTA −3.0800 (0.0023)∗∗∗ −6.0120 (0.0000)∗∗∗ FD AFTA −3.3012 (0.0011)∗∗∗ −3.3141 (0.0010)∗∗∗

EU −2.2510 (0.0240)∗∗ −2.3279 (0.0197)∗∗ NAFTA −1.9599 (0.0481)∗∗ −1.7709 (0.0728)∗

MV AFTA −6.6571 (0.0000)∗∗∗ −6.6551 (0.0000)∗∗∗ EU −5.0762 (0.0000)∗∗∗ −4.8264 (0.0000)∗∗∗

NAFTA −6.2182 (0.0000)∗∗∗ −5.8357 (0.0000)∗∗∗

Note: Figure in the parenthesis are probability values for the test statistic to reject the null of unit root process.

shows that their stock markets are actually getting more segmented from each other. A possible reason for the opposite empirical results is due to the different background of liberalization in EU and NAFTA. The EU’s stock market liberalization generally started a long time ago. In the case of NAFTA however, Mexico’s stock market liber- alization only took place during the period of NAFTA formation (refer to Table 2). As

122 C

.W .H

o oy

Table 7. Estimation results.

Model: SMSI = a + b1TRI + b2TO + b3FD + b4MV + ε (6) where TRI, the Trade Regionalism Index is used to explain the segmentation of trading bloc stock markets captured by SMSI, the Stock Market Segmentation Index; The control variables including Trade Openness (TO), Financial Development (FD) and Market Volatility (MV). The slope coefficient b1 is expected to be positive significant if trade regionalism significantly reduces market segmentation of stock markets among the trading bloc members.

Panel A: Parameter Estimate

Trading Bloc a b1 b2 b3 b4

AFTA 0.0000 (0.0062) 0.4598 (0.9989) 0.0432 (0.1375) 0.0247 (0.0149) −0.0053 (0.0041) EU 0.0000 (0.0034) −0.7969∗∗∗ (0.2113) 0.0961 (0.0973) −0.0185 (0.0133) −0.0100 (0.0046)∗∗ NAFTA 0.0000 (0.0040) 0.6439∗∗ (0.3136) −0.2502 (0.1951) −0.0058 (0.0051) −0.0075 (0.0037)∗∗ Panel B: Explanatory Power and Diagnostic Checking

Trading Bloc R2 Adj R2 LogL AIC SC Q (12) Q (24) Q2 (12) Q2 (24) ARCH-LM

AFTA 0.0810 0.0554 159.7413 −2.0771 −1.9763 14.8150 18.7880 13.3980 15.0260 0.1899 [0.2520] [0.7630] [0.3410] [0.9200] [0.6636]

EU 0.1118 0.0871 243.5705 −3.2023 −3.1015 9.0682 17.3920 12.4470 28.8590 0.1307 [0.6970] [0.8310] [0.4100] [0.2250] [0.7183]

NAFTA 0.0737 0.0480 245.4015 −3.2269 −3.1261 11.7070 24.4460 3.0468 11.3640 0.4921 [0.4690] [0.4360] [0.9950] [0.9860] [0.4841]

Note: Figures in the parenthesis are Newey–West’s (1987) standard errors while figures in the [.] are probability values; Adj R2 is the adjusted R2 ; LogLdenote log likelihood value; AIC and SC are Akaike info criterion and Schwarz criterion, respectively; ∗∗∗ and ∗∗ denote significance at 1% and 5%, respectively.

International Economic Journal 123

a result, along the evolving period of NAFTA, higher trade integration with Canada and the US helps very much to synchronize the diverging pricing of Mexican stocks to the world market, which basically can be represented by its NAFTA partners, the US and Canada.

Another intuitive sense for the result is basically related to the directions of EU financial integration. The integration of the monetary and financial system in the EU basically serves as a catalyst to enhance the regional economic competitiveness so that European firms can compete more successfully in the world market. For this, we see that higher real integration might increase the segmentation of EU stock markets for two reasons. First, when real integration in EU gets higher, only some of the relatively competitive markets benefit. This created capital concentration in a few markets and has caused more pricing divergence in the EU, hence stock market segmentation has increased. Second, the trade creation effects for the EU with the rest of the world might have surpassed the trade creation effects within the EU. Thus, a higher economic (trade) exposure within the EU in a way poses a negative effect on stock market integration.

For the other three control variables, only market volatility shows statistical signif- icance in explaining the SMSI. Market volatility is statistically significant at the 5% significant level in both EU and NAFTA, with an expected negative coefficient. The magnitudes however, are much smaller relative to the loadings on the intra-trade ratio. In fact, of all the explanatory variables, intra-trade ratio has the highest magnitude in the estimated coefficient.

Panel B reports the explanatory power of the model and the diagnostic checking result. The R2 values indicate that model (6) captures about 7% to 11% of the variation in SMSI while for the adjusted R2 the range is only 4% to 8%. The log likelihood, AIC and SC values are reasonable and the estimated models are all free from the correlation problem in the residual, including the ARCH effects.

Conclusion

We investigated the impact of trade regionalism on stock market segmentation within AFTA, EU and NAFTA for the period January 1988 – May 2005. We applied intra-trade ratio and pricing errors from TB-CAPM as the proxies for the respective concept and our empirical model has also been controlled for trade openness, financial development and stock market volatility.

In general, the statistical evidence shows that trade regionalism helps to explain stock market segmentation, but only in the developed-countries dominated trading blocs, i.e. EU and NAFTA. However, the impact of trade regionalism is rather mixed. Real sector integration generally reduces market segmentation in NAFTA but the opposite is true in the EU. Other determinants do not have a significant role in explaining trading-bloc stock market segmentation, except for market volatility. The negative sign of market volatility is expected, and it is consistent with the finding of the study on world stock market integration by Carrieri, Errunza and Hogan (2007).

There are a few caveats to this study. The analysis is based on the assumption that TB- CAPM is the correct pricing model in explaining stock market dynamics in the trading bloc. Another shortcoming is that the intra-trade ratio only captures movements in goods but not services and factors of production, such as capital and labor. Other indicators of trade regionalism can be investigated. Future research can also check on the robustness of the asset pricing model by comparing to other more advanced pricing

124 C. W. Hooy

models in constructing the market segmentation measures. Lastly, other new control variables should be explored given that the explanatory power of the current model still has much room for improvement.

Acknowledgements

The author is a PhD candidate at the University of Malaya. This paper was written while he was a visiting scholar in Columbia Business School, New York. The author would like to thanks Andrew Ang, Charles Jones and participants in the PhD seminar of Columbia Business School for valuable inputs. Special thanks to an anonymous referee of this journal for very helpful comments and suggestions. This paper also received inputs from the participants in Malaysia Finance Association Conference, 8–9 May 2006, Sabah, Malaysia. The author is indebted to Kim-Leng, Goh and Eu-Chye, Tan, and is very grateful for the funding from the University Science Malaysia Fellowship and University of Malaya Short-term Research Grant (F0110/2005B). The usual disclaimer regarding errors and omissions applies.

Notes

1. See the website of WTO at http://www.wto.org/english/tratop_e/region_e/regfac_e.htm. This new develop- ment could be the result of strengthening the fundamentals of the world economy through a highly liberalized multilateral trade system, changing political policy, free flow of capital, and reduction in the costs of trade (see Ethier, 2001). Besides the increase in quantity, the ‘new regionalism’ also features a redundant number of memberships, moderate liberalization and is not geographically bounded.

2. Empirically, a popular way of investigating equity market integration is to examine the direct interaction of stock returns of various markets in terms of their correlations, common trends, news transmission, and volatility spillover effects, in relation to a dominant market, such as Japan, the US and the UK; or among a group of regional markets such as the EMU, Nordic, Asian and Latin regions. This approach is referred to as market linkages and it hinges mainly on the law of one price. Following this popular approach, researchers seldom utilized macro- or micro-oriented variables. The literature in this area hence does not show interest in addressing the likely impact of trade regionalism.

3. We are aware that this indicator may not be the perfect proxy, as the intra-trade ratio captures only one of the four types of freedom in factor movements in the following hierarchy of real integration in regionalism: the free flow of goods in a free trade area and custom union; the free flows of goods, services and labor in a common market; the free flows of goods, services, labor and capital in a monetary union.

4. As noted by Carrieri, Errunzi and Hogan (2007), there is yet any paper that has systematically dealt with the issue of determinants that drives stock market integration. As a result, we adopted their determinant model for market integration.

5. We try to extend the stock market liberalization dummy set provided by Kaminsky and Schmukler (2001) to May 2005, assuming no changes after 2002 (in fact, for our sample, the average stock market liberalization constructed for the EU remains at the maximum score over the whole sample – a fixed value of 3; while AFTA and NAFTA also achieved value 3 after January 1992 and January 1994, respectively), but the loading of the dummy set created a convergence problem in the estimation of model (5). The coefficients for financial liberalization also remain insignificant in our test on model (5) without drift, with positive but very small coefficient magnitudes (except AFTA). As a result, we decided not to include the dummy set in our final model. The result of the above estimation is available upon request.

6. According to Engle and Ng (1993), this simple specification is sufficient for most empirical stock modeling purposes.

7. The unit root tests on the new series are without trend and intercept because both of these have been captured in the model reported in Table 4. In addition, we run the unit root tests on the original series and find that many of them are non-stationary in various possibilities (trend and/or intercept). To conserve space, we are not reporting the result here. The result is available upon request.

8. Besides the contemporaneous relation estimated in model (6), we also estimate the lag effect of TRI on SMSI from lag 1 until lag 12 (holding others constant) to examine how persistent is the trade regionalism effect. The TRI coefficients estimated for EU and NAFTA generally converge to zero with higher lags, with most of the longer lags not being significant. The TRI coefficient for AFTA is more persistent but almost all of the lags are not statistically significant. The results are available upon request.

International Economic Journal 125

References

Adler, M. 1995. Integration of the North American financial markets. Journal of Investing Fall: 81–88. Adler, M., and R. Qi. 2003. Mexico’s integration into the North American capital market. Emerging Market Review

4: 91–120. Akdogan, H. 1992. Behaviour of systematic risk in a regionally integrated model for stock prices. Economics Letters

39: 213–16. Bekaert, G., and C.R Harvey. 2000. Foreign speculator and emerging equity markets. Journal of Finance 55: 565–613. Bakeart, G., C.R. Harvey, and C. Lundblad. 2005. Does financial liberalization spur growth? Journal of Financial

Economics 77: 3–55. Bhagwati, J. 1993. Regionalism and multilateralism: an overview. In New Dimensions in Regional Integration, eds J.

de Melo and A. Panagariya. New York: Cambridge University Press. Black, F. 1972. Capital market equilibrium with restricted borrowing. Journal of Business 45: 444–55. Carrieri, F., V.R. Errunza, and K. Hogan. 2007. Characterizing world market integration through time. Journal of

Financial and Quantitative Analysis, 42, no.2: 915–940. Chan, K.C., B.E. Gup, and M.S. Pan. 1992. An empirical analysis of stock prices in major Asian markets and the United

States. The Financial Review 27: 289–307. Chen, G.M., M. Firth, and O.M. Rui. 2002. Stock market linkages: evidence from Latin America. Journal of Banking

and Finance 26: 1113–41. Choudhry, T. 1996. Interdependence of stock markets: evidence from Europe during the 1920s and 1930s. Applied

Financial Economics 6: 243–49. Corhay, A., A.T. Rad, and J.P. Urbain. 1993. Common stochastic trends in European stock markets. Economics Letters

42: 385–90. Edwards, S., and R. Susmel. 2001. Volatility dependence and contagion in emerging equity markets. Journal of

Development Economics 66: 505–32. Engle, R.F., and V.K. Ng. 1993. Measuring and testing the impact of news on volatility. Journal of Finance 48: 1749–78. Ethier, W.J. 2001. The new regionalism in the Americas: a theoretical framework. North American Journal of Economics

and Finance 12: 159–72. Ewing, B.T., J.E. Payne, and C. Sowell. 1999. NAFTA and North American stock market linkages: an empirical note.

North American Journal of Economics and Finance 10: 443–51. Fama, E.F., and J.D. MacBeth. 1973. Risk, return and equilibrium: empirical tests. Journal of Political Economy 81:

607–36. Frankel, J.A., E. Stein, and S.J. Wei. 1995. Trading blocs and the Americas: the natural, the unnatural, and the super-

natural. Journal of Development Economics 47: 61–96. Fratzscher, M. 2002. Financial market integration in Europe: on the effects of EMU on stock markets. International

Journal of Finance and Economics 7: 165–93. Heaney, R., and V. Hooper. 1999. World, regional and political risk influences upon Asia Pacific equity market returns.

Australian Journal of Management 24: 131–42. Heaney, R., and V. Hooper. 2001. Regionalism, political risk and capital market segmentation in international asset

pricing. Journal of Economic Integration 16: 299–312. Heaney, R., V. Hooper, and M. Jaugietis. 2000. Regional integration of national stock markets. Asian Review of

Accounting 8: 55–65. Heaney, R., V. Hooper, and M. Jaugietis. 2002. Regional integration of stock markets in Latin America. Journal of

Economic Integration 17: 745–60. Hoekman, B.M., and M.M. Kostecki. 2001. The Political Economy of the World Trading System. New York: Oxford. Johnson, R., and L. Soenen. 1993. Stock market reaction to EC economic and monetary integration. European

Management Journal 11: 85–92. Johnson, R., and L. Soenen. 2003. Economic integration and stock market comovement in the Americas. Journal of

Multinational Financial Management 13: 85–100. Johnson, R., J. Lindvall, and L. Soenen. 1994. EC economic and monetary integration: implications for European

equity investors. European Management Journal 12, no. 1: 94–101. Kaminsky, G.L., and S.L. Schmukler. 2001. On booms and crashes: financial liberalization and stock market cycles.

Mimeo. The World Bank. Kaminsky, G.L., and S.L. Schmukler. 2003. Short-run pain, long-run gain: the effects of financed liberalization. NBER

Working paper 9787. Kanas, A. 1998. Volatility spillovers across equity markets: European evidence. Applied Financial Economics 8: 245–56. Korajczyk, R.A. 1996. A measure of stock market integration for developed and emerging markets. World Bank

Economic Review 10: 267–89. Levine, R., and S. Zervos. 1998. Stock markets, banks, and economic growth. American Economic Review 88: 537–58.

126 C. W. Hooy

MacMillan, J. 1993. Does regional integration foster open trade? Economic theory and GATT’s articles XXIV. In Regional Integration and the Global Trading System, eds Kym Anderson and Richard Blackhurst, London: Harvester - Wheatsheaf.

Monadjemi, M.S., and L. Perry. 1996. Share price movements: a study of four OECD countries. Applied Economics Letters 3: 135–7.

Newey, W., and K. West. 1987. A simple positive semi-definite, heteroskedasticity and autocorrelation consistent covariance matrix. Econometrica 55: 703–8.

Ng, T.H. 2002. Stock market linkages in South-East Asia. Asian Economic Journal 16: 353–77. Seabra, F. 2001. A cointegration analysis between Mercosur and international stock markets. Applied Economics Letters

8: 475–8. Soydemir, G. 2000. International transmission mechanism of stock market movements: evidence from emerging equity

markets. Journal of Forecasting 19: 149–76.