2	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

I. INTRODUCTION

Little is known about whether an international investment portfolio covering multiple nations leads to diversifications gains or not — which are by nature short- term, long-term, or both — for investors. In this paper, we explore diversification opportunity for international portfolios with multiple international emerging and frontier equity market indices.

Our study takes the case of investors who reside in any one of the ten fast- growing Emerging and Frontier Asian (EFA) nation and has an investment portfolio comprising of the MSCI of his or her nation, nine MSCIs of other EFA nations, and the S&P 500.

Typically, cointegration technique and an Error Correction Model (ECM) are used to draw out the long-term opportunities and short-run gains, respectively. Thus far, such analysis has mainly been employed for a portfolio comprising typically of two international stock market indices. The 1997-1998 Asian Financial Crisis (AFC) triggered a wave of studies on Asian markets, some of which examined the linkages between Asian stock markets. These studies suggest that the AFC diminished opportunity to diversify within Asia, particularly for those countries most affected by the crisis (see Chiang et al., 2007; Yang, et al., 2003). Recent studies gauging short-term correlations suggest that linkages between selected Asian markets have diminished since the AFC is time varying in nature (Narayan et al., 2014; Phylaktis & Ravazzolo, 2002). Sriananthakumar & Narayan (2015) show a lack of dynamic conditional correlations between Sri Lanka and the neighboring Asian countries. Existing studies on short-term diversification gains are based on pairwise relationships and not on a portfolio of Asian nations as covered in the present paper.

As noted above while there are several studies3 that examine pairwise long- run diversification opportunities within Asian markets, evidence on a portfolio of Asian markets is either limited or unclear. Only three studies provide some clear link between several Asian stock markets. Batareddy et al., (2012) uses rolling and recursive cointegration to find no case for a cointegrating linkage between the emerging stock markets of India, China, South Korea, and Taiwan. Manning (2002) uses the Johansen cointegration approach to find pairwise cointegration for nine Asian markets for the period 1988-1999. Mukherjee and Bose (2008) find evidence in favor of pair-wise cointegration between seven Asian markets over the period January 1999 to June 2005. Other studies examine emerging markets from different regions, including Asia, but with no clear lessons for diversification within Asia (see Auer, 2016; Kenourgios & Padhi, 2012).

Our study contributes to the issue of short and long-run linkages by looking at linkages between stock markets in ten Asian countries – emerging and frontier – for the period 2000-2013. Two key differences between our study and previous studies

3There are many studies that examine integration between emerging and frontier equity markets however we are only focused on studies that are on Asia and use similar technique as us. For a review of stock market integration see Auer (2016); Al

Asad Bin Hoque (2017); Jayasuriya (2011); Kenourgios & Padhi (2012); Mukherjee & Boss (2008); Narayan (2015); Rehman and Kashif (2018); Rehman et al. (2016).

4	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

.10			CHINA
.05
.00
-.05
-.10
2000	2002	2004	2006	2008	2010	2012
.2			INDIA
.1
.0
-.1
-.2
2000	2002	2004	2006	2008	2010	2012
			INDONESIA

.12
.08
.04
.00
-.04
-.08
2000	2002	2004	2006	2008	2010	2012

Figure 1.

Returns of Ten Emerging and Frontier Markets

Monthly returns for the emerging and frontier Asian markets are reported in

the figure from 2000-2013. (Continued)

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	5
Evidence From Ten Asian Countries

.15			KOREA
.10
.05
.00
-.05
-.10
-.15
2000	2002	2004	2006	2008	2010	2012
			MALAYSIA

.2
.1
.0
-.1
-.2
-.3
2000	2002	2004	2006	2008	2010	2012
			PAK

.08
.04
.00
-.04
-.08
.-12
2000	2002	2004	2006	2008	2010	2012

Figure 1.

Returns of Ten Emerging and Frontier Markets

Monthly returns for the emerging and frontier Asian markets are reported in

the figure from 2000-2013. (Continued)

6	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

PHILIPPINES

.2

.1

.0

-.1
-.2
2000	2002	2004	2006	2008	2010	2012
.4			SRI LANKA
.2
.0
-.2
-.4
2000	2002	2004	2006	2008	2010	2012
.08			THAILAND
.04
.0
-.04
-.08

2000

2002

2004

2006

2008

2010

2012

Figure 1.

Returns of Ten Emerging and Frontier Markets

Monthly returns for the emerging and frontier Asian markets are reported in

the figure from 2000-2013. (Continued)

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	7
Evidence From Ten Asian Countries

Table 1 provides some common statistics on the daily returns. Over the fourteen years, the Bangladesh stock market executed the highest mean daily return, while Thailand delivered the lowest. None of the dates corresponding to the maximum or minimum values were the same for any of the stock markets, implying that extreme daily movements did not coincide. Standard deviation of all the Asian returns is below 0.01. The only exception is the Bangladesh stock market which exhibits a deviation of 0.02. The returns series are negatively skewed in the case of Bangladesh, Malaysia, and Philippines while returns of other countries show positive skewness. The coefficient of kurtosis implies leptokurtic distribution for all the markets.

The pairwise unconditional correlations over the period 2000-2013 are significant for most of the cases (Table 1, panel B). Among other pairs, India’s MSCI returns exhibit high correlation with those of Indonesia, Korea, and Malaysia but for the others, low correlation values imply short-term diversification opportunities. Thailand exhibit insignificant correlation with all markets, except for the Korean MSCI.

Table 1.

Descriptive Properties for Daily Return Series

Variables	Bangladesh	China	India	Indonesia	Korea	Malaysia	Pakistan	Philippines	Sri Lanka	Thailand
Panel A: Descriptive Statistics
Mean	0.0048	0.0001	-0.0001	-0.0003	0.0000	-0.0001	-0.0005	0.0001	-0.0006	0.0000
Max.	0.0980	0.0930	0.1180	0.1100	0.1280	0.1920	0.0770	0.1618	0.2970	0.0680
Corresponding date	11.01.11	02.26.07	05.14.04	10.07.08	09.11.01	01.19.10	05.17.02	01.22.01	06.25.09	01.25.05
Min.	-0.0980	-0.0900	-0.1600	-0.0760	-0.1130	-0.1990	-0.0830	-0.1309	-0.3050	-0.0540
Corresponding date	06.29.11	10.22.01	05.15.09	01.22.08	10.29.08	01.20.10	06.23.08	10.27.08	06.26.09	12.30.04
Std. dev.	0.0217	0.0153	0.0156	0.0139	0.0163	0.0100	0.0134	0.0134	0.0135	0.0150
Skew.	-0.0526	0.1791	0.1772	0.7040	0.6512	-0.8369	0.3172	-0.0969	0.1436	0.2127
Kurt.	7.8640	7.5435	10.7886	9.7273	9.2360	115.3645	6.6048	14.7322	163.8707	4.7916
Panel B: Correlations
Bangladesh	1	0.0214	-0.0344*	-0.0280	-0.0698*	-0.0263	0.0161	-0.0011	-0.0430*	0.0264
China		1	0.1669*	0.1777*	0.1841*	0.1651*	0.0476*	0.0124	0.0136	0.0187
India			1	0.3571*	0.3564*	0.2215*	0.0855*	-0.0325*	0.0273	0.0030
Indonesia				1.0000	0.3800*	0.3294*	0.0936*	-0.0490*	0.0448*	-0.0040
Korea					1	0.3210*	0.0695*	-0.0171	0.0444*	0.0339*
Malaysia						1	0.0766*	-0.0501*	0.0219	0.0064
Pakistan							1	-0.0291	0.0217	-0.0264
Philippines								1	-0.0011	0.0078
Sri Lanka									1	0.0085
Thailand										1

8	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

Panel A of this table displays the common statistics for daily MSCI returns of ten Asian countries over the period 2000-2013. Panel B presents the unconditional correlations among MSCI returns of the Asian countries. * denotes level of significance at 5 percent or better.

Next, the panel ADF unit root tests suggests stationarity of the MSCI indices (excluding one country at a time in returns form (Table 2). These results are further confirmed by two other commonly applied panel unit root tests, Im, Pesaran and Shin (IPS, 2003) and Levin, Lin and Chu (LLC, 2002).

Table 2.

Panel Unit Root Test Results

	ADF Statistics	ADF Statistics	IPS Statistics	IPS Statistics	LLC Statistics	LLC Statistics
	At Levels	1st Difference	At Levels	1st Difference	At Levels	1st Difference
Panel A excluding Bangladesh	15.9298	491.306*	-17.2057	-74.6381*	-1.5251	136.693*
Panel B excluding China	38.5234	1475.79*	-14.0044	-79.9007*	-0.2457	58.3201*
Panel C excluding India	10.2899	888.973*	-3.9005	-79.5178*	-0.8306	66.1503*
Panel D excluding Indonesia	9.6028	850.855*	-3.8706	-79.4992*	-1.2118	56.4005*
Panel E excluding Korea	16.2050	1516.43*	-3.7248	-79.7908*	-1.0283	67.6959*
Panel F excluding Malaysia	12.7977	1192.78*	-3.8141	-79.8693	-1.4955	55.1019*
Panel G excluding Pakistan	36.6636	726.083*	-14.9084	-69.0273*	147.991	3286.60*
Panel H excluding Philippine	12.7977	1192.78*	-4.0867	-80.8901*	-1.4204	53.9834*
Panel I excluding Sri Lanka	9.9572	842.694*	-3.4974	-79.7301*	-0.7376	61.3262*
Panel J excluding Thailand	18.1155	664.201*	-3.3243	-77.7970*	-0.8773	74.3484*

This table presents the panel ADF, IPS, and LLC test results for the EFA portfolios (country js) excluding one country at a time. The tests are conducted with a drift and no trend for the levels and returns of MSCI indices.

Table 3, which displays panel Granger causality test results, sheds more light on the relationship between a single Asian stock market against a portfolio of nine other Asian stock markets. Here we examine whether each of the ten MSCI returns Granger causes the rest of the returns in the panel and vice versa, following Dumitrescu & Hurlin (2012) which allows all coefficients to be different across the cross-sections. We find that the results overwhelmingly point to a bidirectional link between one of the ten nation’s MSCI returns and the other nine MSCI returns portfolio. This finding supports our development of the portfolios.

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	9
Evidence From Ten Asian Countries

Table 3.

Panel Causality Tests

	Direction of Causality		EFA Panel
		W HNC	Z HNC	P-Value
		N,T	N,T
	Bangladesh → Country js	4.61389	3.91486	0.000
	Country js → Bangladesh	89.6253	131.291	0.000
	China → Country js	7.76641	8.63843	0.000
	Country js → China	8.81082	10.2033	0.000
	India → Country js	70.3119	102.353	0.000
	Country js → India	14.7253	19.0653	0.000
	Indonesia → Country js	22.5038	30.7201	0.000
	Country js → Indonesia	68.9182	100.265	0.000
	Korea → Country js	21.1862	28.746	0.000
	Country js → Korea	21.7299	29.561	0.000
	Malaysia → Country js	38.8092	55.1512	0.000
	Country js → Malaysia	20.0118	26.9862	0.000
	Pakistan → Country js	40.3835	57.5101	0.000
	Country js → Pakistan	24.3949	33.5537	0.000
	Philippines → Country js	15.5106	20.2420	0.000
	Country js → Philippine	31.7892	44.6329	0.000
	Srilanka → Country js	53.8444	77.6792	0.000
	Country js → Srilanka	17.7773	23.6382	0.000
	Thailand → Country js	24.9953	34.4533	0.000
	Country js → Thailand	92.3677	135.401	0.000

For Granger causality we have used technique proposed by Dumitrescu- Hurlin (2012), allowing all coefficients to be different across cross-sections. Panel causality test is run between ten Asian countries – each time we take returns of a single country vs a panel of nine Asian countries’ returns, excluding that single country. The equation for this panel causality is presented as . Here, Pit is each of the ten Asian nations’ MSCI; Pjt is a portfolio of MSCI of other nine nations. Null hypothesis states no causal relationship for any cross section panels, i.e. Homogeneous Non- Causality hypothesis (HNC). We assume that βi varies across cross sections. In the table presented below, WN,THNC is the average statistics related to Homogeneous Non- Causality hypothesis (HNC). ZN,THNC is the standardized test statistic converging to chi-square distribution with M degrees of freedom. Lag length is 2 and selected as per Akaike Information Criteria (AIC).

10	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

However, our variance decomposition analyses that use unrestricted VAR suggest that each of the nations is minimally affected by the rest of the nine Asian nations (see Table 4). For India, Korea, and Malaysia, the percentage of the variation in their returns explained by the rest of the nine Asian countries range from 0 to 1%; and from 0 to less than 0.5% for each of the other nations. For each of the ten nations, the nine-Asian country panel begins to matter from the second year and explains the variance more over time.

Table 4.

Forecast Error Variance Decomposition Test

Country	1	2	3	4	5	6	7	8	9	10
Bangladesh	0.0000	0.0154	0.0147	0.0221	0.0268	0.0338	0.0407	0.0486	0.0569	0.0658
China	0.0000	0.0186	0.0282	0.0481	0.0698	0.0968	0.1271	0.1612	0.1983	0.2383
India	0.0000	0.4774	0.3855	0.5463	0.5768	0.6796	0.7571	0.8561	0.9529	1.0576
Indonesia	0.0000	0.0000	0.0160	0.0298	0.0587	0.0919	0.1351	0.1845	0.2413	0.3042
Korea	0.0000	0.0348	0.0773	0.1549	0.2496	0.3683	0.5044	0.6578	0.8247	1.0033
Malaysia	0.0000	0.0478	0.1005	0.1981	0.3137	0.4564	0.6158	0.7909	0.9765	1.1694
Pakistan	0.0000	0.0019	0.0064	0.0087	0.0103	0.0113	0.0121	0.0127	0.0132	0.0136
Philippines	0.0000	0.0004	0.0217	0.0377	0.0762	0.1199	0.1777	0.2440	0.3202	0.4043
Sri Lanka	0.0000	0.0471	0.0518	0.0953	0.1275	0.1773	0.2282	0.2883	0.3525	0.4227
Thailand	0.0000	0.0094	0.0219	0.0440	0.0714	0.1059	0.1458	0.1915	0.2423	0.2979

This table highlights percent contribution of the panel of nine countries stock prices in the single country equity prices not included in that panel. The numbers 110 represent time horizon for variance decomposition and ranges from 1 to 10 years’ period. The forecast error variance decomposition is based on the structural VAR model equation of which is Pit=v+ A1 pjt-1+⋯+Ap pjt-p+ uit. Pit is each of the ten Asian nations’ MSCI; Pjt is a portfolio of MSCI of other nine nations, js. Standard errors for this decomposition analysis are generated using recursive-design wild bootstrap.

III. PANEL COINTEGRATION TEST

We begin our empirical analysis for a typical investor in any one of the EFA nations, i, with an investment portfolio comprising of their own national stock market index, nine other EFA market indices, and a developed market index (S&P 500). Of interest to this paper is the following long run relationship for the portfolio conditioned by global sentiments and oil price movements:

(1)

12	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

This table presents results from three cointegration tests, namely Kao (1999), Maddala and Wu (1999), and Pedroni (1999, 2004). of interest is the long-run relationshipdepictedinequation(1):Pit=δ1i+θ1iPjt+θ2iS&P500it+θ3isentiit+θ4iBrentit+μit. Here, Pit is each of the ten Asian nations’ MSCI; Pjt is a portfolio of MSCI of other nine nations; sentiit are global investor sentiments; Brentit is Brent oil price series; and S&P500it is the price index to the US market. * denotes significance at 5 percent or better.

IV. VECTOR ERROR CORRECTION MODEL (VECM)

Here, we estimate the short-run relationship between the variables using the panel VECM model. Of interest is the relationship portrayed here in equation (2):

(2)

All variables from equation (1) appear in equation (2) in first differenced form, represented by ∆. The parameters to be estimated are δ and θs. The Error Correction Term (ECT), which is one lag of the residual from equation (1) if significant and negative, confirms a stable long-run relationship between the variables identified. The ECT is only significant for returns of India, Korea, Malaysia, Pakistan, the Philippines, and Thailand as the dependent variable. This suggests that a stable long-run relationship between returns of countries i and j exists for these six countries. However, this is not so for the returns of Bangladesh, China, Indonesia, and Sri Lanka, suggesting that the long-term investment opportunities in other EFA nations including the S&P 500 may be stronger for the four nations than for other nations (India, Korea, Malaysia, Pakistan, the Philippines, and Thailand).

Evidence of any short-run linkage between returns of country i and js, is rather scant. Only in the case of Korea, Malaysia, and Pakistan do we notice a significant short-term link with the returns of the other Asian nations (or country js) (Table 6). These results imply that short-term gains are limited for the three nations investing in the other EFA markets compared to other Asian nations investing in the EFA markets.

No other variable show a significant link, except S&P 500 returns for Malaysian returns. Our results thus far suggest stable long-run relationship expressed in equation (1) for same EFA nations but rather limited short-run correlations between any one of the ten Asian countries and the other the nine Asian nations’ returns. The link between the EFA nations and S&P 500 is also lacking.

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	13
Evidence From Ten Asian Countries

Table 6.

VECM Results

			Portfolio	Portfolio	Brent	Brent	Sent.	Sent.	SP500	SP500
	Regressors	Intercept				Oil					ECT (-1)
			ret (-1)	ret (-2)	Oil (-1)		(-1)	(-2)	(-1)	(-2)
						(-2)
	Bangladesh	0.2482	0.0007	0.0015	-0.0051	-0.0070	1.2275	5.2269	-0.0013	-0.0016	-0.0003
		(0.2478)	(0.1233)	(0.1233)	(0.2036)	(0.2032)	(72.8270)	(72.7879)	(0.1516)	(0.1516)	(0.0020)
	China	0.0684	0.0684	0.0684	0.0684	0.0684	0.0684	0.0684	0.0684	0.0684	0.0684
		(0.0683)	(0.0353)	(0.0353)	(0.0561)	(0.0560)	(20.0636)	(20.0528)	(0.0418)	(0.0418)	(0.0020)
	India	0.6533	-0.0955	-0.1108	-0.0211	-0.0218	1.8952	15.2825	-0.0179	-0.0088	0.0040*
		(0.6282)	(0.3432)	(0.3432)	(0.5162)	(0.5153)	(184.6449)	(184.5462)	(0.3844)	(0.3844)	(0.0019)
	Indonesia	0.1314	-0.0705	-0.0434	-0.0120	-0.0103	7.1910	-3.8089	-0.0141	-0.0055	0.0004
		(0.1299)	(0.0685)	(0.0685)	(0.1067)	(0.1065)	(38.1728)	(38.1522)	(0.0795)	(0.0795)	(0.0020)
	Korea	0.0603	-0.0819*	-0.0402	0.0078	-0.0001	0.8241	0.9215	0.0017	-0.0205	-0.0057*
		(0.0634)	(0.0331)	(0.0331)	(0.0521)	(0.0520)	(18.6289)	(18.6188)	(0.0388)	(0.0388)	(0.0020)
	Malaysia	-0.0003	-0.0176*	-0.0075	0.0115	0.0081	2.8134	-3.2628	-0.0141*	-0.0076	-0.8037*
		(0.0088)	(0.0046)	(0.0046)	(0.0072)	(0.0072)	(2.5805)	(2.5791)	(0.0054)	(0.0054)	(0.0056)
	Pakistan	0.7677	0.0114*	0.0114*	-0.0115	-0.0328	10.5469	9.2368	-0.0151	-0.0179	0.0058*
		(0.7433)	(0.0065)	(0.0059)	(0.6107)	(0.6097)	(218.4636)	(218.3466)	(0.4547)	(0.4548)	(0.0034)
	Philippine	0.1785	0.1785	0.1785	0.1785	0.1785	0.1785	0.1785	0.1785	0.1785	0.1785*
		(0.1890)	(0.1007)	(0.1008)	(0.1553)	(0.1550)	(55.5395)	(55.5095)	(0.1156)	(0.1156)	(0.0020)
	Sri Lanka	0.1773	-0.0842	-0.0593	0.0056	-0.0341	-2.2664	6.8962	-0.0276	-0.0150	-0.0002
		(0.1764)	(0.0965)	(0.0965)	(0.1449)	(0.1447)	(51.8461)	(51.8183)	(0.1079)	(0.1079)	(0.0020)
	Thailand	-0.0006	0.0000	0.0000	0.0045	0.0003	1.3541	-1.3820	0.0015	0.0018	-0.1389*
		(0.0057)	(0.0002)	(0.0002)	(0.0047)	(0.0047)	(1.6737)	(1.6728)	(0.0035)	(0.0035)	(0.0028)

The table presents the VECM model (2): ∆Pit=δ2i+ θ1i ∑kn=1 ∆Pjt-k + θ2i ∑kn=1

∆BRENTit-k +θ3i ∑kn=1∆SENTI2,it-k +θ3i ∑kn=1 ∆S&P500it-k +δ1i ECTit-1+ϵit. Here, ∆Pit is each

of the ten Asian nations’ MSCI returns; ∆Pjt is a portfolio of MSCI returns of other nine nations; sentiit are global investor sentiments; Brentit is Brent oil price series; and S&P500it is the price index to the US market. These variables appear in first differenced form, represented by ∆, δ, and θs are the parameters to be estimated. The Error Correction Term (ECT) which is one lag of the residual from equation (1), if significant and negative, confirms a stable long-run relationship between the variables identified. Values in parenthesis are standard errors. * denotes level of significance at 5 percent or better.

14	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

V. LONG-RUN REGRESSION RESULTS

We have established two things thus far: (1) that there exists a stable long-run relationship between the returns of six out of the ten Asian nations and the rest of the Asian nations (js); and (2) that the short-run linkages are only present between three out of ten EFA nations and the rest of the Asian nations. Significant short-term and stable long-run relationships between the EFA nations are signs of diminished diversification gains from EFA based investment portfolios. We have found that from EFA based portfolios, long-term portfolio gains are likely to be higher for Bangladesh, China, Indonesia, and Sri Lanka and short-term portfolio are likely to be higher for compared to the other EFAs, India, the Philippines, Thailand, Bangladesh, China, Indonesia, and Sri Lanka. Our short-term finding for Sri Lanka is consistent with Srinanthakumar & Narayan (2015) who find limited conditional correlations between Sri Lanka and the neighboring countries.

Next we look closely at the long-run linkages between the variables by estimating equation (1) using the Dynamic OLS (DOLS) method.6 For this analysis we excluded the four nations that did not show a significant ECT (Bangladesh, China, Indonesia, and Sri Lanka) (see Table 6). For each of the remaining six Asian countries in the first column, the long-run coefficients for the portfolio of returns of country js, or the other nine Asian countries, are displayed in Columns 2 and 3 (Table 7). Note that the long-run results point to a significant and positive link between most of the individual Asian nations and the portfolio of nine nations’ returns. The only exception is Thailand, showing no significant long-run correlation with the other nine Asian countries’ stock markets. Other variables, except the S&P 500, are significant determinants of individual EFA (country i) returns, although EFA returns are almost always more important. Brent is a significant determinant of returns of all six EFA markets examined while global investor sentiments are found to influence returns of Korea, Malaysia, and Thailand.

Table 7.

Long-Run Regression Results

		Pjt	Brent		Sentiments		SP500
DOLS Results
India	0.2583*	2.7464	0.0109*	2.8466	-0.1574	-0.7689	0.0284	0.1645
	-0.0941		-0.0038		-0.2047		-0.1727
Korea	0.1598*	13.0749	0.0087*	16.1053	-0.1596*	-5.4673	-0.0143	-0.5806
	-0.0122		-0.0005		-0.0292		-0.0247
Malaysia	0.1489*	21.6087	0.0062*	18.1076	-0.1016*	-5.4637	0.0098	0.6203
	-0.0069		-0.0003		-0.0186		-0.0158
Pakistan	2.5506*	74.6252	-0.0387*	-18.0649	-0.1005	-0.8672	-0.0134	-0.1356
	-0.0342		-0.0021		-0.1159		-0.0985
Philippine	0.1991*	6.7064	0.0098*	7.9867	0.0213	0.3215	0.0185	0.3322
	-0.0297		-0.0012		-0.0662		-0.0558
Thailand	0.0001	0.0176	-0.0500*	-51.9761	0.9014*	17.1735	0.0348	0.7768
	-0.0003		-0.001		-0.0525		-0.0448

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	15
Evidence From Ten Asian Countries

This table displays the long-run relationships depicted in equation (1): Pit=δ1i+ θ1i Pjt+ θ2i S&P500it+θ3i sentiit+θ4i Brentit+μit. Here, Pit is each of the ten Asian nations’ MSCI; Pjt is a portfolio of MSCI of other nine nations; sentiit are global investor sentiments; Brentit is Brent oil price series; and S&P500it is the price index to the US market. Note that the long-run estimation only relates to the cointegrated relationships depicted in equation (1). * denotes significance at 5 percent or better. Values in parenthesis are standard errors. 3rd, 5th, 7th, and 9th column represent corresponding t-values for long-run results.

VI. S&P 500 AND DEVELOPING STOCK MARKET LINKAGES

Narayan & Rehman (N-R, 2017), a study that most resonates us, use the same set of Emerging and Frontier Asian (EFA) nations over the same period as ours. The authors examined the response of all EFA nations (expressed within one panel) to changes in other variables including developed nations’ stock markets (including the S&P 500). Like the present paper, several control variables were imposed. N-R (2015) did not allocate the EFA markets (country js) as being part of the investment portfolio as we did in the present study. The authors found that the S&P 500 was both a long and short-term predictor of the EFA panel with daily data.

Our finding on the effect of developed market (S&P 500) is different from N-R (2017). We find that S&P 500 is not a predictor of most of the EFA markets in the short or long-run. The only exception is the Malaysian stock market returns in the short-run. The disparate findings may be explained one major difference between the two papers. N-R (2017) did not consider the EFA markets as part of the portfolio investment in their study. Hence in their models, they did not have a variable Pjt in their model. The present paper includes this variable; hence defined portfolio investment opportunities for each individual EFA country (i) against other EFA nations (js) as well as a developed market, the S&P 500. In other words, in the present paper we have explored overseas investment opportunities for an investor based in one of the EFA nations and investing in other EFA nations and the S&P 500.

Taken together, these papers bring to our attention the importance of including both the emerging and developed markets in the portfolio for better diversification gains in the short and long-run.

VII. ACCOUNTING FOR EXCHANGE RATE EFFECTS

We also estimated models inclusive of exchange rates of the country i against the US dollar on the LHS of the equations examined above. The exchange rate data is sourced from the Thomson Reuters Data Stream Financials. These results, presented in the Appendix (Tables A1-A3) are consistent most times, with exchange rate having an insignificant effect in the short and the long-run. However, some divergences is noticed between the new (with exchange rates) and from above old (without exchange rates) results. First, the stable long-run relationship disappears for India and Pakistan in the latest setting, implying that exchange rate movements require hedging for higher long-run investment opportunities in the two South Asian nations. Second, the short-terms results depicted above continue to hold.

16	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

Malaysia is an exception where the effect of other EFAs, Pjt, disappears after being conditioned by exchange rate movements. Third, the long-run association between the stock markets of Korea and Thailand and Pjt is insignificant with the inclusion of exchange rates.

VIII. CONCLUDING REMARKS

Our study explored the feasibility of portfolio investment within Asia for ten emerging and frontier stock markets over the period 2000 to 2013. We found that the unconditional correlations were significant but weak. Nonetheless, the panel Granger causality test suggests the presence of a bi-directional causal relationship between returns for each of the ten Asian nations (country is) and the nine other nations’ (country js) returns.

The long and short-run analyses conditioned to other determinants of Asian returns, such as Brent oil price, global investor sentiments, and S&P 500 returns, revealed several cases of stable long-run linkages but rare cases of short-run linkages between returns of countries is and js.

The long-run relationships between returns of six individual Asian countries and the portfolio of the other nine Asian countries’ returns were positive and significant. These long-run effects were actually stronger than other determinants of stock markets, including Brent oil, global sentiments, and the S&P 500.

In particular, our study suggests that the strongest long and short-term diversification gains for investors in these Asian nations occur if stock market interactions are with Bangladesh, China, Indonesia, and Sri Lanka. Interactions with the other six Asian countries will bring more significant benefits in the short- term than the long-run.

These results imply that diversification gains within Asia are promising. However, whether these gains are comparable to having a portfolio of developed nations or other emerging and frontier markets is left as part a future research agenda.

REFERENCES

Abdullah, A. M., Saiti, B., Masih, M., (2016). The impact of crude oil price on Islamic stock indices of South East Asian countries: Evidence from MGARCH- DCC and wavelet approaches, Borsa Istanbul Review, 16(4), 219-232

Al Asad Bin Hoque, H. (2007). Co-movement of Bangladesh stock market with other markets: Cointegration and error correction approach. Managerial Finance, 33(10), 810-820.

Auer, B. R. (2016). On time-varying predictability of emerging stock market returns. Emerging Markets Review, 27, 1-13.

Baker, M., Wurgler, J., & Yuan, Y. (2012). Global, local, and contagious investor sentiment. Journal of Financial Economics, 104(2), 272-287.

Batareddy, M., Gopalaswamy, A. K., Huang, C-H., 2012. The stability of long-run relationships: A study on Asian emerging and developed stock markets (Japan and US), International Journal of Emerging Markets, 7(1), 31-48.

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	17
Evidence From Ten Asian Countries

Chang, S. C., Chen, S. S., Chou, R. K., & Lin, Y. H. (2012). Local sports sentiment and returns of locally headquartered stocks: A firm-level analysis. Journal of Empirical Finance, 19(3), 309-318.

Chiang, T. C., Jeon, B. N., & Li, H. (2007). Dynamic correlation analysis of financial contagion: Evidence from Asian markets. Journal of International Money and Finance, 26(7), 1206-1228.

De Long, J. B., Shleifer, A., Summers, L. H., & Waldmann, R. J. (1990). Noise trader risk in financial markets. Journal of Political Economy, 98(4), 703-738.

Dumitrescu, E. I., & Hurlin, C. (2012). Testing for Granger non-causality in heterogeneous panels. Economic Modelling, 29(4), 1450-1460.

Jayasuriya, S. A. (2011). Stock market correlations between China and its emerging market neighbors. Emerging Markets Review, 12 (4), 418-431.

Im, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115(1), 53-74.

Kao, C. (1999). Spurious regression and residual-based tests for cointegration in panel data. Journal of econometrics, 90(1), 1-44.

Kenourgios, D. & Padhi, P. (2012). Emerging markets and financial crises: regional, global or isolated shocks? Journal of Multinational Financial Management, 22(1), 24-38.

Lee, C., Shleifer, A., & Thaler, R. H. (1991). Investor sentiment and the closed‐end fund puzzle. The Journal of Finance, 46(1), 75-109.

Levin, A., Lin, C.F., & Chu, C. S. J. (2002). Unit root tests in panel data: asymptotic and finite-sample properties. Journal of Econometrics, 108,1-24.

Lin, C-C., Fang, C-R., & Cheng, H-P., (2014) The Impact of Oil Price Shocks on the Returns in China’s Stock Market, Emerging Markets Finance and Trade, 50(5), 193-205

Maddala, G. S., & Wu, S. (1999). A comparative study of unit root tests with panel data and a new simple test. Oxford Bulletin of Economics and statistics, 61(S1), 631-652.

Manning, N. (2002). Common trends and convergence? South East Asian equity markets, 1988–1999. Journal of International Money and Finance, 21(2), 183-202.

Mukherjee, P. & Bose, S. (2008). Does the stock market in India move with Asia? A multivariate cointegration-vector autoregression approach. Emerging Markets Finance and Trade, 44(5), 5-22.

Narayan, P. K. and S. Narayan (2010). Modelling the Impact of Oil Prices on Vietnam’s Stock Prices. Applied Energy. 87(1), 356-361.

Narayan, S. (2015). Are Asian stock market returns predictable? Emerging Markets Finance and Trade, 51(5), 867-878,

Narayan, S., & Nguyen, T.T., (2015). Does the Trade Gravity Model depend on Trading Partners? Some evidence from Vietnam and her 54 trading partners, International Review of Economics and Finance, 41, 220-237.

Narayan, S. & Smyth, R., (2015), The Financial Econometrics of Price Discovery and Predictability, International Review of Financial Analysis, 42, 380-393.

Narayan, S., & Rehman, M (2017). Diversification opportunities between the Emerging and Frontier Asian (EFA) and Developed Stock Markets, Finance Research Letters, 23, 223-232.

18	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

Narayan, S., Sriananthakumar, S., Islam, S. Z., (2014) Stock Market Integration Emerging Asian economies: Patterns and causes, Economic Modelling, 39, 19-31.

Pedroni, P. (1999). Critical values for cointegration tests in heterogeneous panels with multiple regressors. Oxford Bulletin of Economics and statistics, 61(S1), 653- 670.

Pedroni, P. (2004). Panel cointegration: asymptotic and finite sample properties of pooled time series tests with an application to the PPP hypothesis. Econometric theory, 20(3), 597-625.

Phylaktis, K., & Ravazzolo, F. (2002). Measuring financial and economic integration with equity prices in emerging markets. Journal of International Money and Finance, 21(6), 879-903.

Rehman, M. U. & Kashif, M. (2018). Commonalities between financial and market integration and equity return co-movements in emerging and frontier markets, AESTIMATIO, The IEB International Journal of Finance, 17, pp. 184-203.

Rehman, M. U., & Shah, S. M. A. (2016). Does Bilateral Market and Financial Integration Explains International Co-Movement Patterns1. International Journal of Financial Studies, 4(2), 10.

Singh, P., Kumar, B., & Pandey, A. (2010). Price and volatility spillovers across North American, European and Asian stock markets. International Review of Financial Analysis, 19(1), 55-64.

Sriananthakumar, S., & Narayan, S., (2015). Are prolonged conflict and tension deterrent for stock market integration? The case of Sri Lanka. International Review of Economics and Finance, 39, 504-520.

Yang, J., Kolari, J. W., & Min, I. (2003). Stock market integration and financial crises: the case of Asia. Applied Financial Economics, 13(7), 477-486.

Portfolio Diversification Opportunities Within Emerging and Frontier Stock Markets:	21
Evidence From Ten Asian Countries

Table A3. Long-Run Regression Results

This table displays the long-run relationships depicted in equation (1): Pit=δ1i+ θ1i Pjt+θ2i Brentit+θ3i Sentimentsit+ θ4i S&P500it+ θ5i Exchangeit+μit. Here, Pit is each of the ten Asian nations’ MSCI; Pjt is a portfolio of MSCI of other nine nations; sentiit are global investor sentiments; Brentit is Brent oil price series; Exchangeit is the exchange rate to US dollar; and S&P500it is the price index to the US market. Note that the long-run estimation only relates to the cointegrated relationships depicted in equation (1). * denotes significance at 5 percent or better. Values in parenthesis are standard errors. 3rd, 5th, 7th, 9th and 11th column represent corresponding t-values for long-run results

	Pjt		Brent		Sentiments		SP500		Exchange
DOLS Results
India	-3.2679	-1.1701	0.2117	1.4821	1.7518	0.2220	5.2278	0.7839	-0.0007	-0.0333
	(2.7929)		(0.1428)		(7.8898)		(6.6687)		(0.0215)
Korea	-0.2261	-0.8545	0.0292*	2.1254	-0.0214	-0.0283	0.5031	0.7859	-0.0001	-0.0292
	(0.2646)		(0.0137)		(0.7574)		(0.6401)		(0.0021)
Malaysia	0.1238*	18.3791	0.0067*	19.1516	-0.1111*	-5.7615	0.0177	1.0844	-0.0001	-0.0805
	(0.0067)		(0.0004)		(0.0192)		(0.0163)		(0.0000)
Pakistan	3.4698*	254.3557	0.0134	0.3149	0.0402	0.0171	0.8805	0.4425	-0.0002	-0.0303
	(0.0136)		(0.0426)		(2.3521)		(1.9899)		(0.0064)
Philippine	-0.8936*	-1.0967	0.0670	1.6726	0.5234	0.2264	1.5534	0.7954	-0.0004	-0.0563
	(0.8151)		(0.0418)		(2.3114)		(1.9530)		(0.0063)
Thailand	0.0001	0.0012	-0.0495*	-51.6104	0.8717*	16.4514	0.0421	0.9372	0.0007	4.7072
	(0.0004)		(0.0010)		(0.0530)		(0.0448)		(0.0002)

22	Bulletin of Monetary Economics and Banking, Volume 21, Number 1, July 2018

This page is intentionally left blank