Are the regional Gulf stock markets weak-form efficient as single stock markets and as a regional stock market?

Research in International Business and Finance 33 (2015) 221–246

Contents lists available at ScienceDirect

Research in International Business and Finance j o ur na l ho me pa ge : w w w . e l s e v i e r . c o m / l o c a t e / r i b a f

Are the regional Gulf stock markets weak-form efficient as single stock markets and as a regional stock market? Fouad Jamaani a,∗, Eduardo Roca b a Department of Investment and Finance, College of Finance and Administration, Taif University, Saudi Arabia b Department of Accounting, Finance and Economics, Griffith Business School, Griffith University, Australia

a r t i c l e

i n f o

Article history: Received 30 November 2013 Received in revised form 23 July 2014 Accepted 3 September 2014 Available online 7 October 2014 Keywords: Efficiency Stock market GCC Cointegration Random walk Information asymmetry

a b s t r a c t The purpose of this article is to examine the efficiency of the Gulf Cooperation Council (GCC) stock markets of Saudi Arabia, the United Arab Emirates, Kuwait, Oman, Qatar, and Bahrain. We attempt to answer whether GCC stock markets are weak-form efficient individually or as a group by applying a battery of parametric, nonparametric, unit root, and Johansen cointegration tests to daily index prices denominated in local currencies covering the period December 2003 to January 2013. The findings of our study show that GCC stock markets are not individually weak-form efficient. That is to say, current prices of each GCC stock markets can be predicted from past price changes in that market. The study also finds that collectively, GCC stock markets are not weak-form efficient, as the movements of past prices of one GCC market can be used to predict the current price movement of another GCC stock market. This inefficiency could be due to the weak degree of foreign participation and the high concentration in the banking and financial sectors. Finally, the study suggests a number of policy implications for academics, policy makers and investors, and directions for future research. © 2014 Elsevier B.V. All rights reserved.

∗ Corresponding author. Tel.: +614 11758527. E-mail address: fouad.jamaani@griffithuni.edu.au (F. Jamaani). http://dx.doi.org/10.1016/j.ribaf.2014.09.001 0275-5319/© 2014 Elsevier B.V. All rights reserved.

222

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

1. Introduction Over the last few decades, the economies of developing markets have experienced progressive growth, where the business environment has shown many signs of enhancement, resulting in the listing of new, productive and profitable companies. As a result, the investing in these emerging stock markets has attracted a significant amount of offshore, regional, and local migrant investment capital to consider emerging stock markets as new, lucrative and diversifiable alternative investment channels. However, this rising interest in investment opportunities in developing equity markets, including in the GCC1 stock markets, has raised questions in relation to the efficiency of these stock markets. An empirical investigation of the efficiency of a stock market begins with identifying the economic environment where this market operates. The identification of this economic environment assists in understanding how a characteristic of a particular economy or equity market may impact on its market efficiency. The concept of market efficiency becomes very important when it begins to influence the economic growth of an economy or investment decisions of investors. From an economic viewpoint, lack of market efficiency of a stock market may influence economic growth through affecting factors such as flow of investment capital, cost of capital, and market returns. As an example, the weak-form of stock market efficiency implies that prices paid for stocks should reflect past prices, and accordingly reflect the underlying value of stocks. This results in efficient investment decisions and optimal allocation of financial resources, hence leading to more productive economic activities and investment choices. Efficiency of stock markets also works as a protective mechanism that prevents stock markets from distortions and arbitrage opportunities resulting from the presence of asymmetric information among market participants (Jensen, 1978). From an investment viewpoint, the presence of asymmetric information among market participants can allow informed investors to strategically utilise fundamental information of particular asset groups in a stock market to identify mispriced assets, leading to enhancing their risk-adjusted returns (Malkiel and Fama, 1970). Thus, the concept of market efficiency may be seen as both a tool to measure the degree of a country’s economic development and as a forecasting tool to make excessive returns by utilising stocks’ historical information to identify underpriced stocks. Previous empirical studies concerned with investigating the efficiency of emerging equity markets showed that stock markets in the developing markets are to a significant extent weak-form efficient (Harvey, 1995; Kim and Shamsuddin, 2008). Up to the present time, research concerned with the examination of the efficiency of the GCC stock markets has been limited, produced fragmented results, and only examined the efficiency of GCC stock market from a single market perspective (Ariss et al., 2011; Elango and Hussein, 2008). For this reason, we aim to examine the efficiency of the GCC stock markets, and whether these markets are weak-form efficient either as single markets or as a regional market. Thus, the research question that we attempt to answer is, are GCC stock markets weak-form efficient as single stock markets and as a regional stock market? This study is organised as follows. Section 2 discusses a number of important economic and stock market characteristics for GCC markets and their implications on the efficiency of these markets; in other words, it shows how these characteristics make GCC equity markets an attractive regional environment to examine the efficiency of emerging stock markets. Section 3 briefly reviews the empirical literature. Section 4 presents the methodologies applied, and the data that is used is described in Section 5. Empirical findings are presented in Section 6, followed by discussion of the results in Section 7. A number of policy implications are discussed in Section 8, and the conclusion is made in Section 9. 2. GCC economies and equity markets: brief review We have chosen Gulf stock markets as an attractive laboratory to examine the efficiency of regional emerging equity markets because these markets share common and distinct economic and stock market features that influence their efficiency. GCC state members share similar economic, geographical,

1

Gulf Corporation Council (GCC), namely Saudi Arabia, the United Arab Emirates (UAE), Kuwait, Oman, Qatar, and Bahrain.

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

223

demographic, social, and religious features.2 Gulf member states are rich in natural hydrocarbon resources including oil and gas resources, where in 2011 the combined GCC regional reserves of oil and gas was 33.5% and 21.3% respectively of the proven world reserves.3 These large reserves of oil and gas resources have led the Gulf economies to depend heavily on hydrocarbon industries as a major source of local and regional economic growth. In 2008 the percentage of hydrocarbon products to GCC Gross Domestic Product (GDP), exports, and government revenues was 51.3%, 85.2%, and 88.2% respectively.4 This substantial reliance on these non-renewable natural resources constitutes the first common and distinct feature related to the GCC economies, and may make for serious economic challenges for the GCC region when the values of these natural resources fluctuate or when they become depleted. The problem is that prices of energy, including hydrocarbon products, are by nature quite volatile and do not provide stable and sustainable local and regional economic growth to GCC member states across the long horizon (Bhattacharyya, 2011). GCC policy makers have become aware of these serious economic challenges, and consequently have been aiming for the last two decades to diversify the sources of their economic growth by initiating remedial plans to reduce reliance on oil and gas exports. One of these remedial plans is to effectively reduce over-reliance on oil and gas revenues. GCC governments aim to reduce the contribution of the hydrocarbon industries to GCC nominal GDP from 49% in 2008 to 31% in 2020.5 The supplementary and practical plan set forth by GCC governments is to expand the growth of the GCC stock markets and simultaneously to enhance the efficiency of GCC equity markets, in order to enhance market sentiment leading to the attraction of more investment. This plan aims to provide alternative and sustainable sources for local and regional economic growth (AlKhazali, 2011). The lack of efficiency of GCC stock markets may hinder these long-term growth plans set forth by GCC policy makers, and GCC leaders are interested in enhancing market liquidity, efficiency, and competitiveness within the regional stock markets (Al-Khazali et al., 2006). That is, it is argued that the improvement of stock market liquidity, efficiency, and competitiveness results in lowering the cost of capital, optimising market returns, and increasing the attractions of cross-border capital flows respectively in the GCC markets (Boughanmi, 2008; Neaime, 2002). These benefits may establish an attractive regional investment environment that works as a supplemental boost to reduce the reliance of the GCC region on oil revenues. The inefficiency among GCC markets may act as a disincentive for genuine regional, migrant, and offshore investment capital to flow into GCC markets (Al Janabi et al., 2010). We discuss two distinct stock market characteristics that may indeed play a major role in influencing the efficiency of these regional equity markets. These features include the weak degree of foreign participation in GCC equity markets and the high concentration in banking and financial sectors. Analytically, the former feature incorporates two challenging aspects including the presence of regulatory restrictions on foreign ownership, and the existence of signs of information asymmetry resulting from weak financial market developments in the GCC region. We further analyse the degree of financial market development using three proxies, including the ease of doing business, the transparency of private and public sectors, and the adequacy of digital financial disclosure capacity measures as exhibited in Fig. 1. Before moving to the presentation of these important stock market characteristics of the GCC region, we present distinct intraregional, interregional, and global market capitalisation comparative analyses of GCC equity markets in order to show the size of these Gulf equity markets. From an intraregional perspective, in 2011 the stock market of Saudi Arabia was the dominant one in size, with 49% market

2 On the 25th of May 1981, leaders of the Kingdom of Saudi Arabia, the United Arab Emirates, Kuwait, Qatar, Bahrain, and Oman announced the birth of the Gulf Cooperation Council (GCC). The objective of this union is to effectively establish bridges of co-operation, integration, and inter-connections amongst member states in political affairs, and military, legal, media, security, and economic matters. 3 Oil and Gas figures are obtained from Annual Statistical Bulletin 2012 published by OPEC. 4 GDP, exports and government revenues figures are obtained from report entitled ‘The GCC hydrocarbon sector: big and getting bigger’, published by National Bank of Kuwait in 2011. 5 Figures are obtained from a report entitled ‘The GCC in 2020: Broadening the economy’, published by The Economist Intelligence Unit Limited, 2010.

224

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

Fig. 1. Features of GCC stock markets resulting in inefficiency.

capitalisation of the combined GCC equity market, followed by 18%, 15%, and 13% to the Qatar, Kuwait, and UAE equity markets respectively. The Oman and Bahrain equity markets were the smallest markets in the region, with a percentage of the combined GCC equity market of 3% and 2% respectively in 2011.6 From an interregional viewpoint, GCC stock market capitalisation in 2011 was equivalent to almost 570%, 67%, 44%, and 31% of the market capitalisation size of the North African,7 Scandinavian,8 ASEAN9 and Latin American10 equity markets. Whilst from a global standpoint, combined GCC equity markets accounted only for 4.31% and 1.45% of the U.S and the world equity market capitalisation in 2011.11 The first distinct stock market feature that GCC markets share, that is likely to have a negative influence on the degree of efficiency of these markets, is related to the presence of the weak level of foreign participation in GCC equity markets. In this context, it is presumed that the level of foreign ownership in stock markets has an adverse relationship with the degree of information symmetry. Previous studies argued that foreign investors play an important role in reducing the presence of information asymmetry in equity markets, as they frequently demand more information disclosure, robust accounting and auditing standards, incentive alignments, and better market monitoring mechanisms (Choi et al., 2013; Jiang and Kim, 2004). Hence, it can be said that greater access for foreign investors to local or regional equity markets in the emerging markets, particularly in the GCC market, would lead presumably to greater information symmetry, resulting in more efficient market characteristics. The weak participation of foreign investors in GCC stock markets can be attributed to the presence of regulatory restrictions on foreign ownership. As an illustration of the current regulatory foreign ownership restrictions imposed in GCC markets, for example, foreign investors are only allowed to directly own 49% of listed companies in the UAE, Kuwait, and Bahrain stock markets. While they can own up to 25% and 70% in Qatari and Omani listed equities respectively, only 25% indirect ownership is allowed in the largest GCC stock market of Saudi Arabia via mutual funds, equity swaps, and Exchange Traded Funds (ETFs).12 To some extent these ownership figures may not seem too bad for some GCC markets, but in linking the impact of the presence of information asymmetry with actual ownership figures, a different picture arises. In 2012, for example, foreign investors participated in only 6%, 3.3%, and 28% of listed companies in Kuwait, Saudi Arabia, and Oman, whereas according to equity market

6

Figures for market capitalisation of listed companies in GCC region are obtained from The World Bank Group. North African region includes Egypt, Morocco, and Tunisia only, due to data unavailability of Libya, Algeria, and Mauritania. 8 Scandinavian region includes Finland, Iceland, Norway, Sweden, and Denmark. 9 Association of Southeast Asian Nations (ASEAN) includes Indonesia, Malaysia, The Philippines, Singapore, and Thailand. 10 Latin American markets include Argentina, Brazil, Chile, Colombia, Mexico, and Venezuela. 11 Figures for equity market capitalisation of North African, Scandinavian, ASEAN, Latin American, US markets are obtained from The World Bank Group. 12 Figures are obtained from a report entitled ‘GCC financial markets: Long-term prospects for finance in the Gulf region’, published by Deutsche Bank in 2012. 7

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246 7.0 6.0 5.0 4.0 3.0 2.0 1.0

225 Ease of access to loans 1-7 (best) Venture capital availability 1-7 (best)

Oman

Saudi Arabia

Kuwait

Qatar

Bahrain

United Arab Emirates

GCC

Developing Asia

Enforcment of securities exchange regulations 1-7 (best)

Fig. 2. Financial market developments in the GCC region. Note: Data are hand collected and constructed by authors, and sourced from the annual Global Competitiveness Report published by The World Economic Forum from 2006 to 2013. Global Competitiveness Reports published prior to 2006 do not contain data related to ease of access to loans, venture capital availability, and enforcement of securities exchange regulations, as these proxies were only introduced to the report since 2006 (World Economic Forum, 2014). The ease of access to loans measure assesses to what extent it is easy to obtain a bank loan with only a good business plan and no collateral, and it is scaled from 1 to 7 where 1 = extremely difficult, 7 = extremely easy. The venture capital availability measure assesses to what extent it is easy for entrepreneurs with innovative but risky projects to find venture capital, and it is scaled from 1 to 7 where 1 = extremely difficult, 7 = extremely easy. Ease of access to loans and venture capital availability data are average data from 2006 to 2013 for all GCC economies, excluding Oman and Saudi Arabia where data are only available from 2007 to 2013. For developing Asia economies (34 economies), ease of access to loans and venture capital availability data are average data from 2006 to 2013. The enforcement of securities exchange regulations measure assesses the effectiveness of the regulation and supervision of securities exchanges, and it is scaled from 1 to 7 where 1 = not at all effective, 7 = extremely effective. Data related to enforcement of securities exchange regulations measure are average data from 2007 to 2013 for both GCC economies and the 34 developing Asia economies, as this proxy was introduced in the Global Competitiveness Report in 2007. Developing Asia data are included for comparability purposes.

rules they are able to participate in 49%, 25%, and 70% of these three GCC stock markets respectively.13 The thin foreign ownership is a crucial equity market characteristic that resulted from the regulatory restrictions imposed on foreign ownership that distinguish GCC markets. Thus, we expect a negative relationship between the presence of the weak foreign participation characteristic and the prevailing status of inefficiency among GCC stock markets. The weak participation of foreign investors in GCC stock markets might also be attributed to the relatively low degree of financial market development in the GCC. We provide a descriptive assessment of the degree of financial market development in GCC stock markets by using various proxies related to the ease of doing business, the transparency of private and public sectors, and the adequacy of digital financial disclosure capacity. As exhibited in Fig. 2, we employ two measures to describe the level of GCC financial market development based on ease of access to loans and venture capital availability.14 As well, we discuss the trustworthiness of GCC financial systems using the degree of enforcement of securities exchange regulations.15 On average, GCC financial markets seem to score only moderately in terms of ease of access to loans and venture capital availability: 4.4 and 3.9 out of 7 respectively, with slight variations across GCC countries. As a result of this, it may be quite difficult for entrepreneurs with risky but innovative projects to access venture capital in the region. Despite this feature of the GCC financial system, it nevertheless seems to be more competitive compared to financial systems in developing Asia, where the latter’s ease of access to loans and venture capital availability measures is less than the former by approximately 32% and 26% respectively. The trustworthiness of the GCC financial system, as measured by the extent that securities regulations are enforced, illustrates a better picture compared to developing Asia on average, as it seems that GCC market participants place some trust in the

13 Figures are obtained from a report entitled ‘GCC financial markets: Long-term prospects for finance in the Gulf region’, published by Deutsche Bank in 2012. 14 Data related to ease of access to loans and venture capital availability sourced from Global Competitiveness Report published by The World Economic Forum. The Global Competitiveness Network has published reports measuring country competitiveness since 1979, reaching global coverage to 148 economies by 2013. The data used in the report are sourced from leading international sources as well as from the World Economic Forum’s annual Executive Opinion Survey, a unique source that captures the perspectives of more than 13,000 thousand business leaders on topics related to national competitiveness (World Economic Forum, 2014). 15 Data related to enforcement of securities exchange regulations sourced from Global Competitiveness Report published by The World Economic Forum (World Economic Forum, 2014).

226

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

10.00 9.00 8.00 7.00 6.00 5.00 4.00 3.00 2.00 1.00 0.00

Transparency of government policymaking 1-7 (best) Favoritism in decisions of government officials1-7 (best)

Oman

Saudi Arabia

Kuwait

Qatar

Bahrain United Arab Emirates

GCC

Developing Asia

Strength of investor protection 1-10(best)

Fig. 3. Transparency Environments of GCC public and private sectors. Note: Data are hand collected and constructed by the authors, and sourced from the annual Global Competitiveness Report published by The World Economic Forum from 2006 to 2013. Global Competitiveness Reports published prior to 2006 do not contain data related to transparency of government policymaking, favouritism in decisions of government officials, and strength of investor protection, as these proxies were only introduced to the report since 2006 (World Economic Forum, 2014). The transparency of government policymaking measure assesses to what extent it is easy for businesses to obtain information about changes in government policies and regulations affecting their activities, and it is scaled from 1 to 7 where 1 = extremely difficult, 7 = extremely easy. The favouritism in decisions of government officials measure assesses the extent government officials show favouritism to well-connected firms and individuals when deciding upon policies and contracts, and it scaled from 1 = always show favouritism, 7 = never show favouritism. Data related to transparency of government policymaking and favouritism in decisions of government officials measures for Kuwait, Oman, and Saudi Arabia economies are only average data from 2007 to 2013, as 2006 data are missing from the report. For developing Asia economies (34 economies), transparency of government policymaking and favouritism in decisions of government officials data are average data from 2006 to 2013. The strength of investor protection index assesses the strength of minority shareholder protections against directors’ misuse of corporate assets for personal gain. The index incorporates three dimensions of investor protections: transparency of related party transactions (extent of disclosure index), liability for self-dealing (extent of director liability index) and shareholders’ ability to sue officers and directors for misconduct (ease of shareholder suits index). The strength of investor protection index is the average of the extent of disclosure index, the extent of director liability index and the ease of shareholder suits index. The index ranges from 0 to 10, with higher values indicating more investor protection. Data related to the strength of investor protection index are reported as average values from 2006 to 2013 for both all GCC economies and the 34 developing Asia economies. Developing Asia data are included for comparability purposes.

enforcement mechanism of their securities regulations while Asian investors display 20% less confidence in theirs. Thus, it can be said that, on average, while there are some difficulties in the ease of conducting business in the GCC region, it is slightly easier than in developing Asia. Apart from the slightly difficult environment for doing business that foreign investors may expect to experience in the GCC region, the transparency environment of GCC public and private sectors seems to be at question and possibly to be the source of information asymmetry, as shown in Fig. 3. We employ two indices to measure the GCC public sector’s transparency which relate to the transparency of government policymaking and favouritism in decisions of government officials,16 while we use an index on the strength investor protection17 to gauge the transparency of the GCC private sector. On average, we obtained a score of 4.76 out of 7 with regards to transparency of the government policymaking and 4.30 out of 7 in relation to favouritism. Based on these scores, it can be said that it is to some extent not difficult for well-connected individuals and businesses to obtain information about related government policy changes and to jump the line when it comes to policies and contracts. A severe transparency problem is notable in the Kuwait public sector, which is almost similar to what is expected in developing Asia compared to other GCC public sectors. As a result, the weak transparency of the GCC public sector seems to spill over to the GCC private sector, as indicated by the moderately weak result of 5.4 out of 10, on average, in the strength of the investor protection index. This implies that it is quite expected that minority shareholders’ rights will be to some extent misused for personal gain, the ability to file lawsuits against directors’ misconduct to be slightly inferior, and financial disclosure rights to be violated. The investor protection climate is fragile for both the GCC and developing Asian regions, where the presence of information asymmetry is a prevalent market feature in developing stock markets (Abdmoulah, 2010; Klapper and Love, 2004). Therefore, the existence of weak

16 Data related to transparency of government policymaking and favouritism in decisions of government officials sourced from Global Competitiveness Report published by The World Economic Forum (World Economic Forum, 2014). 17 Data related to strength of investor protection index obtained from Global Competitiveness Report published by The World Economic (World Economic Forum, 2014).

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

227

transparency to some extent in both the public and private sectors in the GCC markets can facilitate the presence of weak financial market development resulting from the presence of asymmetric information between individuals and businesses. The weak financial market development is also a crucial equity market characteristic that results from the presence of a partly difficult business environment and weak transparent market mechanisms in both private and public sectors that distinguish the GCC markets. These non-friendly investment characteristics would deter foreign investors from exercising active participation in GCC stock markets. Therefore, we expect a negative relationship between the presence of this weak foreign participation characteristic and the information efficiency among GCC stock markets. The third financial market development measure that is likely to have an influence on the degree of foreign participation in these markets, is related to the presence of information asymmetry caused by the inadequate digital financial disclosure capacity of GCC equity markets. Equity markets in the Gulf region have a short history compared to the global financial markets. Most of the GCC stock markets were established and regulated in the 1980s and 1990s,18 and only recently applied electronic trading platforms. The electronic trading incorporates the setting up of an official trading website for every stock market or every listed company that provides frequent release of importantly relevant historical and fundamental information. As electronic trading has been implemented in most GCC equity markets for little more than a decade, it is not unreasonable to assume a linkage between the unavailability of digital financial information and the presence of information asymmetry in GCC markets. This linkage is already established and empirically examined in the literature, in the sense that more use of digital means for information disclosure of financial information of listed firms is expected to have a negative impact on the presence of information asymmetry in equity markets (Healy and Palepu, 2001). To put this linkage to the test from a GCC stock market perspective, Ismail (2002) empirically investigated the use by 128 GCC listed companies of Internet-based information disclosure, and found that almost 61% of these firms do not have websites to publish their financial information to investors, leading to the constitution of information asymmetry. Similar findings are emphasised by Joshi and Al-Modhahki (2003) and Hussainey and Al-Nodel (2009). It is repeatedly argued in the literature that the presence of information asymmetry in equity markets has negative implications on the weak-form efficiency status of markets, so that inefficiency of such markets is expected (Peress, 2010). Since it is known that digital means of communicating financial information to GCC investors is a specific feature of GCC stock markets, we expect the presence of inefficient market characteristics in GCC equity markets, driven by the presence of information asymmetry, in turn caused by the lack of effective dissemination of digital financial information. Further scrutiny into the composition of individual and collective GCC equity markets provides the second distinct shared characteristic of these regional equity markets that may have implications for their degree of information efficiency. In 2012, banking and financial services dominated almost 50% of the indices of the Qatar, Kuwait, Bahrain, Oman, and the Abu Dhabi Securities Market.19 By contrast, banking and financial services accounted for 13% and 23.9% in the Saudi Stock Exchange and Dubai Financial Market respectively.20 From the combined GCC perspective, banking and financial services constituted almost 41% of the composite of the combined Gulf equity markets in 2012. GCC banking and financial sectors have well-established regional and global banking and financial networks (Al-Khazali et al., 2006; Neaime, 2012, 2002). These global and regional banking and financial inter-linkages may have implications on the efficiency status of GCC stock markets. It is claimed that financial markets that are tight in international, interregional, and intraregional financial linkages are more crisis-prone than other unconnected markets (Bekaert et al., 2005; Neaime, 2005). These global and regional financial

18 The Kuwait Stock Exchange (KSE) was the first regulated GCC market in 1983, and the UAE’s was the last regulated GCC equity market in 2000. The Saudi and Kuwaiti stock markets are the first two markets that traded shares electronically, where the latter commenced electronic trading in 1995 and the former in 1990. The Qatar, UAE, Bahrain, and Oman stock markets introduced electronic trading in 2002, 2000, 1999, and 1998 respectively. Figures are obtained from a report entitled ‘GCC Economic Outlook’ published by Global Investment House in 2011. 19 Figures are obtained from Abu Dhabi Securities Exchange, Bahrain Bourse, Kuwait Stock Exchange, Muscat Securities Market, and The Qatar Exchange official websites. 20 Figures are obtained from Dubai Financial Market and Saudi Stock Exchange official websites.

228

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

linkages can work as facilitating vehicles to ease the transfer of external financial shocks across regional banking (Guyot et al., 2014). The argument is that a banking crisis in one country gets transmitted to other countries via cross-country banking linkages (Lagoarde-Segot and Lucey, 2006; Sandoval Junior and Franca, 2012). In this context, banks play a significant role in the development of financial crises, and previous empirical studies have shown that a financial crisis has an adverse effect on the informational efficiency of stock markets (Azad, 2009; Risso, 2008). As GCC stock markets have a high concentration in banking and financial sectors, and experienced two financial crises in 2006 and 2008, we expect the current efficiency status of these equity markets to be inefficient. The analysis of these GCC stock market characteristics is very important, as it shows that the overreliance of GCC economies on oil and gas revenues constitutes a serious economic challenge, as the prices of these resources are volatile and have limited reserves. GCC stock markets have only one path to follow forward, namely enhancing market efficiency, if they are keen to attract regional and global investment capital to their equity markets. The problem is that our analysis of the regional stock markets shows that these markets possess characteristics that are likely to lead to informationally inefficient markets. First, there is thin foreign participation in GCC equity markets, disincentivised by the presence of regulatory restrictions on foreign ownership and weak financial market developments. Second, there is high concentration of GCC equity markets in banking and financial business that is regionally and globally well connected which may serve as channels for the spill-over of excessive intraregional, interregional, and global volatility into these markets during times of financial crises which will then have adverse effects on market efficiency. Hence, given the presence of these characteristics among GCC stock markets, we are more inclined to anticipate that these markets are likely to fail weak-form efficiency tests, meaning that present stock price changes are not independent of past price movements. 3. Literature review Although the empirical literature of the weak-form efficiency of the GCC stock markets is still at its infancy stage, what exists has produced fragmented results and highlighted a number of issues related to problematic methodological limitations of previous studies that aimed to test the behaviour of GCC stock market returns. One of the earliest empirical studies that examined the efficiency of GCC stock markets used daily index data for the Kuwaiti stock market from 1975 to 1987. The study rejected the normality hypothesis and employed parametric autocorrelation and nonparametric Runs tests to examine the random behaviour of the Kuwaiti stock returns and concluded the presence of significant dependency of current prices changes to past price changes (Gandhi et al., 1980). Another study extended the sample of the previous research to include the Saudi and Kuwaiti stock markets for a similarly short data frame from 1985 to 1988. The study found that stock returns are not normally distributed and used autocorrelation and Runs tests. In a similar conclusion to the previous study, the weak-form efficiency is rejected due to the presence of significant predictability of the behaviour of current prices movements to past price movements in these markets (Butler and Malaikah, 1992). Further evidence arguing for the rejection of the random behaviour of GCC stock market returns is provided by a study that used index prices of Saudi Arabia, UAE, Kuwait, Oman, Qatar, and Bahrain from 2001 to 2006 using the Runs test. On the other hand, the randomness of Bahraini and UAE stock market returns was suggested using autocorrelation and Runs tests for daily index prices from 1996 to 2000 and from 2001 to 2003 for the first and second study respectively. Both studies found that both Bahraini and UAE stock returns are not normally distributed (Moustafa, 2004; Rao and Shankaraiah, 2003). Other GCC studies employed the parametric variance ratio (VR) test suggested by Lo and MacKinlay (1988) and produced fragmented results. On the one hand, the random walk behaviour of Saudi, Kuwaiti, Omani, and Bahraini stock market returns was examined using weekly index prices from 1994 to 1998 through using a number of tests including the parametric autocorrelation and VR test (Dahel and Laabas, 1999). The study found the market returns of these markets are not normally distributed and the results of the autocorrelation and VR tests supported the random walk behaviour. On the other hand, the efficiency of the Saudi stock markets was investigated using Runs and VR tests for daily index prices from 2003 to 2004 and provided statistically significant rejection of the

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

229

weak-form efficiency of the Saudi market returns (Onour, 2004). Another study used VR and Runs tests to examine the randomness of daily sectorial indices data between 2000 and 2005 for the Dubai Financial Market (DFM) and the Abu Dhabi Securities Market (ADSM) (Squalli, 2006). The study found that UAE stock markets are not weak-form efficient except for the banking sector in the DFM and the insurance sector in the ADSM. In regard to the different results achieved by these previous empirical studies about the status of the weak-form efficiency of GCC stock markets, they seem to share two methodological problems that may have led to these conflicting results. First, a critical limitation to parametric tests including autocorrelation and VR tests is that they require time series to be normally distributed. This normality requirement implies that employing autocorrelation tests to investigate the dependency of stock prices, in the case that normality of stock prices is rejected, may constitute statistical bias to the results (Guidi and Gupta, 2013). This statistical bias may result in a high likelihood of misjudgement of the null hypothesis of random walk. Second, it has been argued in the literature that although the nonparametric Runs test does not have a normality requirement, it does suffer from different types of departure from random behaviour, consequently rejecting the random walk hypothesis due to the use of shorter sampling periods (Elango and Hussein, 2008). As the abovementioned empirical GCC studies employed short sampling periods, on average less than four years, some inaccuracy in their results may be expected. However, it is argued that the randomness of stock prices may not be tested by parametric and nonparametric dependency tests alone, as randomness of stock prices requires also the condition of non-stationarity (Guidi and Gupta, 2013). As randomness of stock markets requires the existence of non-stationarity amongst time series, previous GCC studies used unit root tests to examine the condition of non-stationarity among GCC stock market returns. The concept behind unit root tests is to examine whether or not a time series has a unit root. Accepting the null hypothesis of a unit root implies the non-stationarity of the time series that rejects the possibility of the presence of a predictable component in time series. Similarly to the parametric and nonparametric tests, the results of the empirical studies that aimed to test the weak-form efficiency of GCC stock returns using the unit root application yielded fragmented results. For example, early studies concerned with examining the weak-form efficiency of the Gulf stock markets used unit root tests for weekly index prices for the Bahrain, Kuwait, Oman, and Saudi Arabia stock markets from 1994 to 1998. The study showed that unit root tests accepted the randomness of the Gulf stock markets (Dahel and Laabas, 1999). On the other hand, the vast majority of the studies that used unit root tests to investigate the dependency of successive price changes of the GCC stock prices rejected the null hypothesis of independency. For instance, the randomness of the Saudi Arabia stock prices was examined using unit root tests and daily stock prices from March 2003 to June 2006. The results of the unit root tests showed that the Saudi stock market is not weak-form efficient (Onour, 2009). Other studies used unit root tests for studying the weak-form efficiency of the GCC stock markets using different frequency of data, and concluded with supportive evidence rejecting the randomness hypothesis of the GCC stock market returns (Al Ashikh, 2012; Bley, 2011; Lee et al., 2010). Apart from these fragmentary results of unit root tests, it has been argued in the literature that unit root cannot stand by itself to provide an adequate judgement of the random walk, as the process of unit root can have predictable components (Azad et al., 2014). This implies that unit root can only examine the stationarity status of time series, without further ability to show if these stationarity time series are serially correlated or not. The process of random walk implies that equity prices must not be correlated, for which unit root tests cannot provide any information. The empirical investigation of the efficiency of the GCC stock markets encountered macroeconomic challenges to account for the impact of financial crises, including the GCC price bubble crisis in 2006, and the global financial crisis in 2008, that are likely to cause structural breaks in stock market returns behaviour. Lee et al. (2010) investigated whether the EMH holds in a group of developed and developing stock markets, including the Saudi Arabia stock market, under various economic developments. The study accounted for the impact of the presence of two structural breaks that resulted from the twin hits of the 1998 Asian crisis and the 2006 GCC bubble crisis. The results of the study concluded that the Saudi stock market does not follow random walk behaviour. In addition, Bley (2011) examined the weak-form efficiency of the GCC stock markets taking into consideration the five-year sub-periods from 2000 to 2009, in which he accounted for the impact of the GCC price bubble crisis in

230

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

2006 and the Global Financial Crisis in 2008. Consistent with the results of Lee et al. (2010), the daily and weekly results of Bley (2011) rejected the weak-form efficiency of GCC stock markets. Further challenging issues relating to the characteristics of GCC stock market prices are identified in the literature, including the impact of thin trading on the accuracy of the results of the weak-form efficiency tests. Previous studies argued that GCC stock markets, like most of the emerging markets, are characterised by infrequent trading behaviour (Bekaert et al., 2005; Harvey, 1995; Moustafa, 2004). Thin trading happens when equities do not trade at every consecutive interval, consequently the infrequent trading effect can yield statistical biases in the time series of stock prices, leading to induced serial correlations (AlKhazali, 2011). Some of the previous GCC empirical studies concerned with examining the efficiency of GCC stock markets recorded different results before and after correcting for the effect of thin trading. For example, VR and Runs tests were used to examine the null hypothesis of random walk of the GCC stock markets using weekly index returns from 1992 to 1998. The study showed different results before and after the consideration of the effect of thin trading (Abraham et al., 2002). The study revealed that GCC markets tested as not efficient before correcting for infrequent trading, yet showed the opposite results after accounting for infrequent trading effects. These results are consistent with the results of some previous GCC studies (Al-Khazali et al., 2007; AlKhazali, 2011). A recent study, on the other hand, investigated the efficiency of the Gulf stock markets using different parametric and nonparametric tests on daily and weekly market returns from 1999 to 2010 (Al-Ajmi and Kim, 2012). The study found no differences in results before and after correcting for infrequent trading, although the study argued that less signs of random walk were recorded after correction for thin trading with weekly data. This view of the marginal difference of results before and after correcting for the infrequent trading effect argued that differences in the dataset is the cause of previous studies observing an influential difference after correcting for the thin trading effect. For example, GCC studies that showed different results before and after correcting for thin trading used stock index prices data ranged from 1992 to 2007, in which trading volumes in the Gulf stock markets were indeed very thin. The argument that these authors identified is that during the 1990s and the first half of the 2000s, the trading volume of most of the GCC stock markets was drastically low, but since the second half of the 2000s trading volume in the Gulf stock markets has multiplied dramatically. Thus, once trading volumes are no longer thin, correction for infrequent trading becomes obsolete, as empirically proven by Al-Ajmi and Kim (2012). To this point in the GCC empirical literature, the weak-form efficiency of the GCC stock markets was only investigated from a single perspective. The investigation of the weak-form efficiency of the emerging stock markets has been extended to consider testing the collective weak-form efficiency of a group of stock markets using the application of cointegration testing (Guidi and Gupta, 2013). The existence of cointegration has implications on the efficiency of equity markets. It has been debated by previous studies that if the prices of two stock markets are cointegrated, then the movement of one market can be predicted from the past prices movements of the other (Granger, 1986; Lence and Falk, 2005). In this sense, these two markets cannot be weak-form efficient as investors can use past price changes in one market to predict the current price movement in another market. Recently, the weak-form efficiency of the ASEAN stock markets, including those of Indonesia, Malaysia, the Philippines, Singapore, Thailand and Vietnam, was examined using daily index prices from 2000 to 2011. The study used a number of unit root, parametric, and nonparametric tests to investigate the weak-form efficiency of the ASEAN stock markets from an individual perspective. The researchers also used the Johansen cointegration test to examine the weak-form efficiency of these Asian stock markets from a collective perspective (Guidi and Gupta, 2013). The study showed that Indonesia, Malaysia, the Philippines and Vietnam are not weak-form efficient, whereas the Singapore and Thailand stock markets are weak-form efficient individually. From a collective perspective, the results of the cointegration test rejected the null hypothesis of cointegration among the ASEAN stock markets. This indicates that although these markets are not weak-form efficient individually, they are efficient collectively. From an economic perspective, the results imply that there are no common shared trends among the ASEAN stock prices that can be exploited to predict the movement of other ASEAN stock market prices. The results of this study may impose challenges related to the individual and collective efficiency

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

231

characteristics of regional stock markets as previously suggested by Lim et al. (2008), to the extent that regionally close economies may in fact not share similar stock market efficiency characteristics, as initially theorised. The presence of cointegration amongst individual stock markets in the GCC region would imply that Gulf stock markets are not efficient as a group, as the movement of prices in an individual GCC market can be used to predict the movement of other GCC stock market prices. To the best of our knowledge, using the cointegration test to examine the collective efficiency of the GCC stock markets has not been applied by previous studies on the efficiency of the Gulf equity markets. Thus, we aim to fill this gap in the empirical literature by using the application of Johansen cointegration in order to add a new perspective to the testing methods of the efficiency of the GCC equity markets. This leads us to the second part of our research question, are GCC stock markets weak-form efficient as a regional stock market?21 It is evident from the review of the previous empirical literature related to the examination of the weak-form efficiency of the GCC stock markets that mixed results have been achieved through using different methods and different data sets. Our study aims to extend the empirical literature of the market efficiency in emerging markets in four ways. First, we focus on the regional GCC stock markets that have similar economic reliance on oil and gas revenues. The lack of stock market efficiency among GCC markets is likely to be a disincentive for regional, migrant and offshore investment capital to invest in GCC markets, consequently leading to continuing reliance on the non-renewable oil resources to boost local and regional economic growth. An empirical investigation of the current efficiency status of the GCC stock markets would inform GCC leaders about the current efficiency condition of their equity markets, so that they can implement the appropriate strategy based on their agenda. Second, regional and international investors would benefit from having current empirical results about the returns profitability and efficiency status of GCC stock markets. For example, identifying a GCC stock market that has positive market return rates and simultaneously is inefficient, may imply the presence of a valuable investment opportunity to generate consistently excessive returns by using technical and fundamental information of that stock market. This possibility of such investment opportunities may require retesting of the current efficiency condition of GCC stock markets. Third, unlike previous GCC studies, our study considers the use of current and longer daily index prices data of the Gulf equity markets from the end of December 2003 to the end of January 2013, in order to provide a broader and more accurate view about the current expected market returns and the current status of the efficiency of these markets. It has been identified in the literature that high frequency data such as daily data produces more robust results (Boehmer and Wu, 2013; Lo and MacKinlay, 1988). Fourth, our review of the empirical literature showed that a single testing method seems to be inadequate to draw accurate conclusions about the status of weak-form efficiency of a stock market, due to the inherent limitations suffered by every model. We acknowledge this limitation and consider the employment of a wide range of independency tests, including a number of unit root, parametric, nonparametric, and Johansen cointegration tests, in order to maximise the robustness of our results. The application of the cointegration test has never been used previously in the examination of the weak-form efficiency in the Gulf stock markets from a collective perspective. Hence, the methodology that will be employed by our study extends the horizon of the ways previous studies have focused on investigating the weak-form efficiency of the Gulf equity markets from a regional perspective. 4. Methodology The procedure applied to test the random walk behaviour of the GCC stock markets is chosen on the basis of the implications of EMH suggested by Malkiel and Fama (1970). This procedure implies that if all relevantly available information is fully reflected in equity prices, then: Successive price changes will be normally distributed, hence stock returns are normally distributed (Condition 1). Also, successive price changes will be independent, thus no serial correlations will exist between

21

Our research question is, are GCC stock markets weak-form efficient as single stock markets and as a regional stock market?

232

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

stock returns (Condition 2). To test the normality of the GCC stock market returns, Condition (1), the null hypothesis is as follows: Ho: GCC stock markets’ returns are normally distributed. The reason we first test for normality is that if time series are not normally distributed, then the results of parametric tests such as VR and multiple variance ratio (MVR) tests should be used with caution, as these tests are designed to test dependency of time series for normally distributed time series. The results of tests that employ parametric tests in the cases when time series are not normally distributed may result in statistical errors (Kim and Shamsuddin, 2008). We acknowledge this potential limitation of the parametric tests; so, if time series tested to be non-normally distributed, then we use non-parametric tests such as Wright and Runs tests, as the structural properties of these nonparametric tests do not require time series to be normally distributed. Therefore, we employ two types of commonly applied normality tests, including tests of skewness and kurtosis, and the Jarque–Bera test to examine the normality of the GCC stock market returns. In order to investigate and test for randomness, Condition (2), the study examines null and alternative hypotheses to test the serial dependence of the returns behaviour of the GCC stock markets. Accordingly, the null hypothesis is as follows: Ho: GCC stock markets’ returns are independent as single stock markets, thus are weak-form efficient individually. To test the randomness of returns in the GCC stock markets, we follow similar testing techniques to those applied by Granger (1986), Fifield and Jetty (2008), Guidi and Gupta (2013), and Mobarek and Fiorante (2014). Hence, we apply a range of tests, including Augmented Dickey–Fuller (ADF), Phillips–Perron (PP), VR, MVR, Wright, Runs, and Johansen cointegration, in order to examine whether GCC stock markets are weak-form efficient as individuals or as a group. 4.1. Augmented Dickey–Fuller (ADF) test The unit root test is designed to test for stationarity of time series, as the presence of nonstationarity indicates the existence of randomness that supports the weak-form efficiency hypothesis (Azad, 2009). We employ three unit root tests including ADF and PP tests. The ADF test is carried out to find out if the time series being analysed is stationary. When using ADF, one should consider whether to include exogenous variables in the test regression. For example, there is the choice of choosing among two models, including a constant (Eq. (1)), and a constant and a linear time trend (Eq. (2)), as shown below. Model two :

yt−1 = c0 + ıyt−1 + ˇ

p 

yt−1 + t .

(1)

i−1

Model three :

yt = c0 + c1 t + ıyt−1 + ˇ

p 

yt−1 + t .

(2)

i−1

yt is a series that follows an autoregressive (AR) process. c0 and c1 are optional exogenous regressors, ı and ˇ are parameters to be estimated. t is assumed to be white noise. The null hypothesis is the presence of a unit root, so not rejecting that hypothesis means the series follows a random walk. The hypothesis is evaluated using the conventional t-ratio for ı: ta =

ıˆ ˆ se(ı)

.

(3)

ˆ is the coefficient standard error. where ıˆ is the estimate of ı, and se (ı) 4.2. Phillips–Perron (PP) test The stationarity of GCC market returns can be tested using the procedure developed by Phillips and Perron (1988) as a non-parametric alternative and controlling test for serial correlations while

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

233

testing the unit root. Phillips and Perron (1988) proposed to estimate the long-run variance by the Newey and West estimator. The difference between the PP and the ADF tests is the manner in which heteroskedasticity and serial correlation in errors are dealt with. The PP test’s idea is to use non Augmented Dickey–Fuller regression, then make adjustment for bias which may exist because of correlation in innovation terms. The PP test is non-parametric and its specifications are as follows: Pt =  + ıPt−1 + εt .



Pt =  + ˇ t −

 1

2

T

(4) + ˛Pt−1 + εt .

(5)

in the equations, Pt is natural price index logarithm during time t while  represents a constant. Also ˇ and ˛ are parameters which need to be estimated, and εt happens to be an error term. Eq.  (5) has  the constant term only, while Eq. (4) has both a constant and linear trend terms,  and ˇ t − 12 T . The PP tests hypotheses may be stated as H0 if the series has unit root and H1 if it is stationary or does not have unit root. 4.3. Variance ratio (VR) test Lo and MacKinlay (1988) provide two different test statistics for the random walk properties of a series with different sets of null hypotheses. First, they make a strong assumption that the residuals are independently and identically distributed (IID) like Gaussian with variance  2 , while normality assumption is considered as not strictly necessary, called a homoskedastic random walk hypothesis. Alternatively, they outline a heteroskedastic random walk hypothesis to allow for more general forms of conditional heteroskedasticity and dependence, termed as the martingale hypothesis, offering a set of sufficient conditions for the errors to be a martingale difference sequence. The test is based on the assumption of linearity that the variance of a random-walk term is a linear function over time, which has made the test the most commonly applied procedure to test the hypothesis of random walk. The VR statistic is given as follows: VR(q) =

 2 (q) .  2 (1)

(7)

1 (YT − YT −q − q)2 . Tq

(8)

where T

 2 (q) =

t=1

and the estimator of variance should be adjusted for bias when T is replaced with (T − q + 1) or (T − q + 1)(1 − q/T). Thus the variance ratio statistic is given as: z(q) =

VR(q) − 1 [ˆs2 (q)]

−1/2

.

(9)

the statistic is normally distributed for an appropriate choice of variance at q. The variance is given under the normality assumption by: sˆ2 (q) =

2(2q − 1)(q − 1) . 3q1

(10)

and under the martingale difference sequence it is given by:

sˆ2 (q) =

q−1    2(q − j) j−1

q

2

· ıˆ j .

(11)

234

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

where the ıˆ j is given by:

 T

2

t=J+1

(yt−j − ) ˆ (yt − ) ˆ

 T

ıˆ j =

(y − ) ˆ t=J+1 t−j

2

2





.

(12)

Hence, the variance ratio statistic is based on the underlying distribution to be normal if the series and errors of the series to be IID Gaussian. z(q) statistic under the homoskedastic and heteroskedastic approaches to examine the null hypothesis that VR(q) = 1. If the results of the test show that: VR(q) < 1.

VR(q) > 1 or

(13)

then understudy stock market returns are said to be positively and negatively serially correlated respectively. 4.4. Multiple variance ratio (MVR) test The main difference between the VR and MVR tests is that the latter overcomes the limitation of single interval testing of the former, to include multiple interval testing (Chiang et al., 2010). Due to commonly applied statistics at differently selected values of q, the VR conditionality is valid for each difference in q > 1. To hold down the size of the joint test, Chow and Denning (1993) test provides a statistic examining the maximum absolute value of multiple variance ratio statistics. The p-value for the Chow–Denning statistic is bounded from the VR test by the probability for the studentized maximum modulus (SMM) distribution for a set of m given parameters and T degrees-of-freedom. MVR test provides a procedure for the multiple comparison of the set of variance ratio estimates with unity. The null hypothesis of the VR is as follows: VR(q) = 1 thus MVR(q) = VR(q) − 1 = 0.

(14)

under the random walk null hypothesis, MVR can be tested as follows: H0i : MVR(qi) = 0 or H1i : MVR(qi) = / 0 for any

i = 1, 2, . . ..m.

(15)

4.5. Runs test The Runs test is a nonparametric test used to examine the independence of successive price changes. A run is defined as the successive occurrences of the same pattern of changes; positive changes in consecutives runs, negative changes in consecutives runs, and so forth (Mobarek and Fiorante, 2014). The Runs test applies for the randomness of the runs of series of stock market prices or returns. For a measure of random stock market returns or prices, the R runs should be nearly equal to the expected number of m runs. The null hypothesis of Runs test is thus the weak-form efficiency of GCC stock market returns. The calculation of the expected number of runs can be achieved by applying Eq. (16), m as: m=

N(N + 1) − N

3

n2 i=1 i

.

(16)

where N is the total number of runs, i is the number of positive and negative changes in series and n is the total changes in category of change. The expected number of runs is approximately normally distributed for n > 30 for standard deviation equal to  m (ibid). The runs are calculated by following equation:

3  m =

i=1



n2i + N(N + 1) − 2N N 2 (N − 1)

3

n3 i=1 i

− N3

1/2 .

(17)

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

235

where the standard normal distribution for conducting Runs test can be determined from the following equation: Z=

R − m ± 0.5 ∼N(0, 1). m

(18)

the number 0.5 in the above specification determines the continuity adjustment being negative when R ≥ m and positive for R < m. Butler and Malaikah (1992) have indicated if the total number of runs is exceeding (falling behind) the expected number of runs, the occurrences of positive (negative) Z statistic is obtained. It is also to be noted that a negative Z value determines the positive serial correlation, while a positive Z value highlights the negative serial correlation. If the value of Z statistics is above 1.96 at 5% level of significance, the non-randomness is violated. 4.6. Wright test Wright (2000) provides nonparametric VR tests based on ranks (R1 and R2) that can be more powerful than the tests suggested by Lo and MacKinlay (1988) in conditions where the distribution of the given series is not normal. The rank statistic is given by: rit =

r(Yt ) −



T +1 2

.

(19)

(T −1)(T +1) 12

r2t = −1

 r(Y )  t

T +1

.

(20)

in case where the r is tied, the denominators can be modified to account for the ties. Based on the above the test statistics can be derived as:



R1 (k) =

 R2 (k) =

1 Tk

1 Tk

T

t=k (r1t 1 T

+ . . . + r1t−k+1 )2



T

T t=k

r2 t=1 1t

(r2t + · · · + r2t−k+1 )2 1 T

T

r2 t=1 2t

r2t = F − 1(r(DYt)(T + 1).

× ∅(k)−1/2 .

(21)

 × ∅(k)

−1/2

.

(22) (23)

where T shows the number of observations of first differences of variables Y, (stock prices), ∅ is the asymptotic variance, r is the rank of Y among at 1, T, and ˚−1 is the inverse of the standard normal cumulative distribution function. 4.7. Johansen cointegration test To this point, all previously mentioned tests have the ability to examine only the random walk behaviour or the weak-form efficiency of GCC stock markets from the individual perspective. As the aim of this paper is to also investigate the weak-form efficiency of the GCC stock markets from a collective viewpoint, we use the application of cointegration testing. Thus, the null hypothesis is as follows: Ho: GCC stock markets are weak-form efficient as a group; thus, there is no long-term cointegration between GCC stock prices. The presence of cointegration among GCC stock market prices has implications for the EMH. Previous studies debated that if two stock market prices are cointegrated, then past movements of one set of market prices can be used to predict the movement of other market prices (Granger, 1986; Guidi and Gupta, 2013). The basic concept behind cointegration testing is that for two or more nonstationary time series, there may be a long-term relationship existing among them. This concept implies

236

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

that although the two variables might move randomly from each other, they might converge in the long run, resulting in what is called a cointegrating relationship (Guidi and Gupta, 2013). The presence of cointegration among stock market returns would imply that the movement of one market could be used to predict the movement of the other. Thus, the presence of cointegration among equity markets would produce supporting evidence leading to the refutation of the weak-form efficiency. The cointegration test is based on maximum likelihood estimation that proposes two distinct tests for determining likelihood ratio, including the trace and maximum Eigenvalue tests. The trace test: determines r cointegrating vectors’ null hypothesis alongside the substitute n cointegrating vectors’ hypothesis. If the value of r is 0, it therefore implies that a relationship does not exist among the nonstationary variables, hence no cointegration exists. Maximum Eigenvalue test: determines r cointegrating vectors’ null hypothesis alongside alternative hypothesis of (r + 1) cointegrating vectors. The Johansen test adopts a starting point from a vector autoregression of the order p represented as: yt =  + A1 yt−1 + · · · + Ap yt−p + εt .

(24)

where yt represents n × 1 integrated variables’ vector generally represented as I(1) while εt represents an n × 1 innovations vector. The two likelihood ratio tests include trace test and maximum Eigenvalue test, and are shown in the following two equations respectively. Jtrace = −T

n 

ˆ i ). ln(l − ␭

(25)

t−r+1

ˆ r+1 ). JMaxim = −T ln(l − ␭

(26)

ˆ i shows the ith biggest canonical correlation. In the equations, T shows the size of the sample while ␭ The advantage associated with the model is that it can be used in estimation of several cointegration relationships. 5. Data and descriptive statistics Our study comprises daily stock market index prices of Saudi Arabia, Dubai, Abu Dhabi, Kuwait, Oman, Qatar, and Bahrain stock markets in local currencies obtained from Thomson Datastream from the end of December 2003 to the end of January 2013. The rationale behind choosing daily data frequency has been identified in the literature to the effect that different or inferior results may be expected when employing low frequency data such as yearly, quarterly, monthly and weekly data, as compared to high frequency data such as daily data (Lo and MacKinlay, 1988). One interpretation of this situation is that as the lag length augments in daily stock prices data, correlation becomes smaller. Hence, this phenomenon clearly argues the notion that the larger the interval of the observations of equity prices, the less significant is the lag price in determining the future stock price changes (Boehmer and Wu, 2013). The selection of our data range from December 2003 to the end of January 2013 has two motivations. As we identified previously in the GCC stock market section, two financial hits struck GCC markets in 2006 and 2008. Firstly, investigating the current efficiency status of the GCC stock markets using a current GCC index data to capture the pre-crisis and post-crisis effect on the efficiency of GCC equity markets is necessary. In this paper, we are not going to examine the changes of GCC stock market efficiency pre- and post-crisis, rather examining the entire independence of successive price changes of GCC stock returns from 2003 to 2013. Secondly, this data range is the longest and most current dataset that we can obtain from Thomson Datastream. The testing process of daily raw GCC stock prices data are transformed into log returns22 data using the following formula: (Rt) =

Ln(Pt) (Pt − 1)

(27)

22 The aim of using GCC log returns instead of index prices is to convert the index price data into continuously compounded rates, which is the more common applied practice than employing discrete compounding.

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

237

Table 1 Summary statistics for daily stock market returns.

Oman Saudi Arabia Kuwait Qatar Bahrain Abu Dhabi Dubai

No of observations

Mean

Maximum

Minimum

SD

Skewness

Kurtosis

2397 2397 2397 2397 2397 2397 2397

0.000329 0.000191 0.000021 0.000321 −0.000079 0.000221 0.000270

0.080388 0.163995 0.074561 0.094220 0.036132 0.398183 0.102199

−0.086990 −0.116816 −0.075200 −0.093592 −0.049200 −0.364920 −0.121573

0.011044 0.017647 0.010485 0.015499 0.006064 0.016963 0.018312

−0.887619 −0.566263 −0.272469 −0.373149 −0.431974 1.156443 −0.126897

18.18 13.38 10.44 09.56 09.22 220.11 08.66

Jacque– Bera 23,343 10,882 55,540 43,530 39,333 47,084 32,080

P-values 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

where Rt = GCC market returns for period, Ln = natural log of GCC stock prices, Pt = GCC index price at period t, and Pt-1 = GCC index price at period t − 1. On average, daily market returns for all GCC stock markets are positive with the exception of the Bahraini stock market. The highest and lowest daily positive market returns were recorded in Oman (.000326%) and Kuwait (.000021%) respectively, as shown in Table 1. The lowest and highest observed daily standard deviations (SDs) were recorded in Bahrain (.00606%) and in Dubai (.018312%) respectively. On average, risk-return tradeoff seems to not hold very well in GCC stock market returns, for example, as the highest daily SD of (.018312%) in Dubai is not achieved by the highest recorded daily mean return of (.000326%) in Oman. Daily kurtosis values of all GCC index returns are higher than three, suggesting that returns distributions are with fatter tails. The skewness values for all daily GCC index returns are negative with the exception of Abu Dhabi index returns. These negative values indicate that asymmetric tail extends further towards negative returns than positive returns, suggesting early signs of non-normality of GCC stock returns. This non-normality suggestion is further confirmed by all daily GCC stock returns, as all Jarque–Bera test statistic values exceed the value of 5.99 of which the normality hypothesis of GCC stock returns is rejected at 5% level of significance. 6. Results of the study 6.1. Results of unit root tests We examine the presence of a unit root in GCC stock returns using ADF and PP tests with both constant, and constant and trend. The examination of the presence of a unit root in the GCC stock markets implies the non-stationarity behaviour of the individual GCC stock returns, thus proving the randomness of GCC stock returns. We can observe that the values of the test statistics for the ADF and PP under both models are considerably higher than the critical values of Mackinnon equivalent critical values, as shown in Table 2. The results of the daily ADF and PP tests show that all GCC stock market

Table 2 Unit root tests for daily stock market returns. ADF Constant Oman Saudi Arabia Kuwait Qatar Bahrain Abu Dhabi Dubai

−40.44612 −45.92687* −42.66640* −39.13006* −41.96402* −53.60351* −31.81637* *

PP Constant and trend

Constant

−40.50217 −45.94655* −42.71700* −39.14652* −42.20579* −53.63229* −31.91387*

−40.06483 −40.06483* −42.6628* −39.14043* −43.32189* −53.48788* −49.16496*

*

Constant and trend *

−40.12973* −45.95849* −42.69838* −39.13724* −43.01435* −53.55762* −49.06730*

* Indicates rejection of the null hypothesis of random walk at (−3.432) 1% and (−3.961) 1% level of significance for models with constant and for the model with constant and trend respectively.

238

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

Table 3 VR tests for daily GCC stock market returns. Number of days, Q, in holding period Oman VR (q) Homoscedasticity (q) Heteroscedasticity-robust (q) Saudi Arabia VR (q) Homoscedasticity (q) Heteroscedasticity-robust (q) Kuwait VR (q) Homoscedasticity (q) Heteroscedasticity-robust (q) Qatar VR (q) Homoscedasticity (q) Heteroscedasticity-robust (q) Bahrain VR (q) Homoscedasticity (q) Heteroscedasticity-robust (q) Abu Dhabi VR(q) Homoscedasticity (q) Heteroscedasticity-robust (q) Dubai VR (q) Homoscedasticity (q) Heteroscedasticity-robust (q) *

Q=2

Q=4

Q=8

Q = 16

0.60982 −19.0989* −5.466594*

0.330076 −17.52811* −5.480285*

0.152722 −14.02057* −4.907736*

0.07155 −10.32478* −4.174539*

0.549661 −22.0436* −9.422655*

0.254851 −19.49631* −8.919774*

0.129994 −14.39666* −7.024946*

0.065177 −10.39566* −5.446313*

0.573773 −20.86337* −9.515027*

0.294258 −18.46526* −9.457908*

0.143619 −14.17118* −8.094956*

0.071982 −10.31998* −6.251655*

0.617088 −18.74316* −8.330987*

0.336404 −17.36253* −7.925499*

0.154484 −13.9914* −6.879076*

0.078615 −10.24623* −5.571194*

0.570458 −21.02563* −11.44112*

0.29278 −18.50393* −11.1157*

0.142272 −14.19348* −9.60664*

0.074332 −10.29385* −7.683775*

0.452586 −26.79534* −2.062029**

0.233901 −20.04446* −1.873771**

0.113952 −14.66212* −1.811281**

0.056999 −10.4866* −1.744885**

0.482882 −25.31236* −11.95115*

0.24705 −19.70042* −9.891465*

0.121099 −14.54386* −8.145314*

0.065832 −10.38838* −6.540587*

, and ** Indicate rejection of the null hypothesis of random walk at (2.59) 1% and (1.69) 5% level of significance.

returns reject the null hypothesis of the presence of a unit root, at 1% significance level for constant (−3.432) and constant and trend (−3.961). 6.2. Results of variance ration (VR) test The null hypothesis of independence of successive price changes of the GCC stock market returns is further investigated by VR tests. In order to accept the null hypothesis of random walk of the GCC stock markets, consequently accepting the weak-form efficiency, the VR test should equal a value of one in order to reject the presence of autocorrelation between lags. If the results of VR test go below or above one, then positive or negative autocorrelations imply the possibility of predicting future price changes. The presence of such potential for exploitation would provide evidence towards rejecting the random walk hypothesis of the GCC stock markets, leading to the rejection of the weak-form efficiency of these Gulf markets. Table 3 displays the results of VR tests for daily GCC stock market returns at different time lags, including 2, 4, 8, and 16 lags, with the consideration of homoscedastic and heteroscedastic growth of GCC stock returns’ residuals. Results of the daily VR tests for individual GCC stock markets are quite similar, as all VR results for all lags are above the critical value of 2.59 with negative signs indicating the rejection of random walk at 1% level of significance. Under the homoscedasticity assumption, all GCC stock market returns illustrate significantly negative serial correlations in all lags at 1% level of significance (indicated with * signs). However, even after heteroscedastic growth is assumed, signs of significant autocorrelations are evident for all lags and for all GCC stock market returns at 1% level of significance (indicated with * signs). Hence, we show that the null hypothesis of independent successive price changes of all daily GCC stock market returns is significantly rejected at 1% level of significance, indicating the rejection to the null hypothesis of the weak-form efficiency of the GCC stock markets.

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

239

Table 4 MVR Tests for the GCC stock markets returns. Daily MVR2

MVR1 Oman Saudi Arabia Kuwait Qatar Bahrain Abu Dhabi Dubai

*

19.09890 22.0436* 20.86337* 18.74316* 21.02563* 26.79534* 25.31236*

5.480285* 9.422655* 9.515027* 8.330987* 11.44112* 2.062029 11.95115*

MV1 is the homoscedastic and MV2 is the heteroscedastic-robust version of the Chow–Denning test. * Indicates rejection of the null hypothesis of random walk at (3.089) 1% level of significance.

6.3. Results of multiple variance ratio (MVR) test Table 4 presents the results of the MVR tests for daily stock returns for the GCC stock markets, under homoscedastic and heteroscedastic growth assumptions of GCC stock returns. According to Table 4, signs of significant autocorrelation coefficient prevail for all GCC stock market returns and critical values of Cow–Denning of 2.68 (5%) are exceeded for all GCC stock returns. For example, under the homoscedastic and heteroscedastic growth assumptions for daily GCC stock returns, the null hypothesis is significantly rejected at 1% level of significance for all markets due to the presence of significant positive autocorrelation effect (indicated with * signs). Although the MVR test overcame the individual testing problems of the VR test, both the VR and MVR are prone to statistical bias that can be caused by the violation of the normality prerequisites. However, in the literature review section, we highlighted that parametric tests, including VR, and MVR tests, are likely to cause a statistical bias from the use of non-normally distributed time series, as they all have normality prerequisites. To overcome the problems of non-normality while testing for dependency of the GCC stock market returns, we use nonparametric tests, including Wright and Runs tests. 6.4. Results of Wright test The results of R1 and R2 for Wright test, including 2, 5, 10, and 10 time lags for daily GCC stock returns, are reported in Table 5. The results of R1 and R2 for the GCC stock returns for all lags exhibit significant serial correlation coefficient values indicating the relation between past and current stock price movements. The possible predictability, as illustrated by the significant R1 and R2 results, supports the rejection of the random walk hypothesis of the GCC stock markets at 5% statistical level of significance. Thus, we can state that GCC daily stock market returns cannot be weak-form efficient. 6.5. Results of Runs test The guideline of accepting the null hypothesis of the independence of successive price changes of the GCC stock returns requires a close match between the expected and actual number of runs. The significant presence of large or low numbers of actual runs compared to the number of expected runs would provide rejection evidence against the random walk hypothesis. From a statistical viewpoint, the null hypothesis of random walk of the GCC stock market returns is rejected at 5% level of significance when Z value of the Runs test is higher than the critical value of 1.96. Negative Z values indicate the presence of significant autocorrelation among stock returns. Table 6 displays the daily results of Runs tests for the GCC stock market returns using the median (Z) and mean (Z*) of returns as a base for estimating the number of runs. It should be noted that we use Z* just for results comparability rather than for concluding results. All reported values of Runs tests for both Z and Z* results are higher than the critical value of 1.96. This indicates a significant rejection of the randomness of the daily GCC stock returns caused by the presence of statistically significant serial correlation.

240

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

Table 5 Wright’s nonparametric VR tests using ranks for daily GCC stock market returns. Number of K lags

K=2

K=5

K = 10

K = 30

Oman R1 R2

−17.61628** −18.45808**

−15.28699** −16.19739**

−11.33996** −12.03307**

−6.683448** −7.092166**

Saudi Arabia R1 R2

−22.33245** −22.98358**

−16.46014** −17.18421**

−11.74368** −12.3987**

−6.907769** −7.254416**

Kuwait R1 R2

−19.34092** −20.26477**

−15.70807** −16.68921**

−11.54924** −12.3429**

−6.792431** −7.22876**

Qatar R1 R2

−15.19351** −16.78736**

−14.2316** −15.63567**

−10.45506** −11.67984**

−6.077322** −6.89871**

Bahrain R1 R2

−20.01641** −20.94608**

−15.81231** −16.82346**

−11.78683** −12.58638**

−7.028656** −7.437216**

Abu Dhabi R1 R2

−17.16343** −18.50601**

−14.87896** −15.99237**

−10.96826** −11.7555**

−6.453498** −6.915291**

Dubai R1 R2

−23.12857** −24.74983**

−16.66604** −17.64975**

−11.99045** −12.70467**

−6.849078** −7.355725**

**

Indicates the rejection at 5% level of significance.

6.6. Results of Johansen cointegration test As all previously applied tests are only able to examine the weak-form efficiency of the GCC stock markets from an individual perspective, we implement the Johansen cointegration test in order to examine the possibility of the presence of long-term relationships among GCC stock prices. The Johansen cointegration test requires stock prices to be integrated at the same order to be able to examine the likelihood of observing stationary cointegrating relationship among non-stationary stock prices. In order to satisfy this essential requirement, Table 7 presents the results of daily ADF and PP tests for the GCC stock prices. As most of the GCC economies and stock markets have witnessed progressive growth over the last few years, intercept and trend are included in the testing equation for ADF and PP tests at level and at first difference in order to account for this growth. The results of both ADF and PP tests collectively show that all GCC stock prices are non-stationary at level, while all are stationary at first difference (log returns) for daily stock prices. The non-stationarity of GCC stock prices at level for all ADF and PP values of all GCC stock market prices are below −3.961 (1%), as displayed in Table 7. These results imply that all GCC stock market prices are integrated at the same

Table 6 Runs test for daily GCC stock market returns.

Oman Saudi Arabia Kuwait Qatar Bahrain Abu Dhabi Dubai **

Observations (N)

n (+)

n (−)

n (0)

Expected runs (m)

Actual runs (R)

Z

Z*

2397 2397 2397 2397 2397 2397 2397

1461 1199 1424 1244 1265 1307 1291

936 1198 973 1198 1153 1090 1106

0 0 0 0 0 0 0

1142 1199 1157 1199 1198 1190 1192

929 1112 1039 1002 1061 975 1105

−9.14** −3.58** −5.00** −8.07** −5.60** −8.84** −3.59**

−6.87** −4.00** −5.99** −7.81** −5.03** −7.44** −2.30**

Indicates the rejection at (1.96) 5% level of significance.

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

241

Table 7 ADF and PP tests for testing the order of integration of the daily GCC stock market prices. PP test

ADF test Daily

Level

First difference

Level

First difference

Oman Saudi Arabia Kuwait Qatar Bahrain Abu Dhabi Dubai

−1.512118 −2.350622 −1.598868 −2.613338 −2.316542 −2.331444 −2.172475

−40.50217 −45.94655* −42.717* −39.14652* −42.20579* −53.63229* −31.91387*

−1.47592 −2.33794 −1.62154 −2.4807 −2.22617 −2.32265 −2.21434

−40.12973* −45.95849* −42.69838* −39.13724* −43.01435* −53.55762* −49.0673*

*

* Indicates rejection of the null hypothesis of random walk at (−3.961) 1% level of significance for the model with constant and trend.

Table 8 Results of Johansen cointegration test for GCC stock prices. Daily GCC stock prices

Model 1:Intercept without trend

Model 2: Intercept with trend

Number of cointegrating vector

Trace statistic

5% critical values

MaxEigen statistic

5% critical values

Trace statistic

5% critical values

MaxEigen statistic

5% critical values

None At most 1 At most 2 At most 3

170.2 109.1 66.7 36.8

125.6** 95.7** 69.8 47.8

61.1 42.3 29.8 20.5

46.2** 40.0** 33.8 27.5

201 138 85.5 49.9

150.5** 117.7** 88.8 63.8

62.6 52.6 35.6 22

50.5** 44.4** 38 32

**

Indicates the rejection at 5% level of significance.

level (first difference) at 1% level of significance, meeting the prerequisite requirement of the Johansen cointegration test. Before choosing the type of the deterministic trend assumptions required for the Johansen cointegration test, the number of lags is determined to be 4 lags in order to allow for longer testing windows. Among the five available models23 that the Johansen cointegration test provides, we assume the presence of a linear deterministic trend in the GCC stock prices. The logic behind this assumption is that as progressive economic and stock market growth is observed in the GCC region, linear growth is expected. However, under this deterministic trend assumption, there are two options, including the consideration of including trend, or trend and intercept, in the cointegrating equation. In order to observe whether the inclusion or exclusion of time trend in the cointegrating equation may cause differences in the results, Table 8 presents the results of the Johansen cointegration test for both intercept without and with trend. All daily GCC stock price data for both models with and without trend, both Trace and Max-Eigen test statistics rejected the null hypothesis of the presence of no cointegration between GCC stock prices. Two cointegrating vectors are present for daily GCC stock prices at 5% level of significance, (indicated with ** signs) in Table 8. These results imply the presence of strong long-term common trends among GCC stock prices, indicating that GCC stock markets are not weak-form efficient as a group. From an investment viewpoint, GCC investors may be able to predict, for example, future movement of Dubai stocks from the past movement patterns of the Saudi or Qatari stock markets. The proven inefficient stock markets characteristics would imply that other regionally close economies like the GCC may indeed share similar efficiency characteristics. The presence of long-term relationships of the GCC stock market prices using Johansen cointegration tests was acknowledged by previous scholars, including Abraham and

23 The first model has not intercept and trend VAR, while the second model permits intercept without trend in the Cointegrating Equation (CE) and no intercept in VAR. The third model permits intercept in the CE and VAR, while the fourth model permits intercept in CE and VAR along with linear trend in CE but not for VAR. The fifth model permits intercept and quartic trend in CE and VAR.

242

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

Seyyed (2006), Al-Khazali et al. (2006), Hammoudeh and Choi (2006), and Neaime (2002). However, to the best of our knowledge, none of these authors or other studies has used the application of the cointegration analysis of stock market prices to examine the collective weak-form efficiency in the GCC region. Thus, the results of this study extend the horizon of the literature of GCC stock market efficiency, in which the long-term relationship of stock prices can be considered as a tool for testing the weak-form efficiency of a group of stock markets in the Middle East, and elsewhere. 7. Discussion of the results The prevailing significant serial correlation among GCC stock market returns and the long-term price co-movements may be caused by the presence of excessive stock market volatility and information asymmetry. In fact, the results are in line and are a consequent response to the stock market efficiency characteristics that GCC markets share, and were outlined in the economic and stock market section. This excessive market volatility found its way through GCC stock markets due to a specific stock market characteristic that GCC markets together share. The high concentration of GCC equity markets in banking and financial services, coupled with well-developed global and interregional financial linkages, may be acting as a mobiliser of this volatility. Previous studies argued that global and interregional connected banking systems play a crucial role in the development of financial crises as financial intermediaries which contribute to the efficient transfer of funds from the cash surplus to the cash deficit (Lagoarde-Segot and Lucey, 2006; Neaime, 2012). It is also well documented in the literature that excessive volatility normally results from financial crisis shocks, and these crises eventually result in reducing the efficiency of stock markets through the augmentation of serial correlation trends within and amongst market returns (Azad, 2009; Risso, 2008). The previous two studies empirically identified the negative relationship of stock market efficiency and financial crises. As GCC stock markets received two financial hits in 2006 and 2008, and their indices are heavily weighted in banking and financial sectors, we can say that the characteristics of the GCC stock markets, coupled with the impact of these two major events that occurred to Gulf equity markets, helped in constituting the current weak-form inefficiency of single and collective GCC stock markets. We believe that the results of our study are very important for four reasons. First, the comprehensive analysis we carried forward from the economic and stock market overview of GCC countries allowed us to reveal some distinct features of these regional equity markets. The rapid growth in GCC stock market trading volumes from the early to late 2000s highlighted the presence and absence of thin trading effects during the early and late 2000s respectively. The results of our study constitute a challenge to the results of Dahel and Laabas (1999), Moustafa (2004), and Rao and Shankaraiah (2003). These studies accepted the efficiency of GCC stock markets using a short range of data, less than four years on average, and their data covered periods where GCC stock markets were characterised as thinly trading markets as empirically proven by Al-Khazali et al. (2007) and Elango and Hussein (2008). In contrast, our study employed a longer and current range of data from 2003 to 2013, the first GCC study that uses such a long and current dataset. The argument we have carried forward from the literature review is that the impact of thin trading becomes obsolete as GCC stock market trading volumes multiplied many times since the mid-2000s. Bley (2011) used a GCC data range from 2000 to 2009, and Al-Ajmi and Kim (2012) employed GCC index prices from 1999 to 2010, and they both corrected for the impact of thin trading. They both found no different results before and after correcting for the effect of infrequent trading, an outcome that our own results reinforce as highly valid and reliable. Secondly, due to the use of current GCC stock market data that captured the full financial effect of the recent financial crises on GCC market returns, our results reveal that not all GCC stock markets produced positive returns as previously thought. The Bahrain stock market produced negative stock market mean returns. Acknowledging the significant effect of the recent credit crisis on equity market returns and particularly banking sector returns, it is no wonder that Bahrain stock market returns produced negative returns as, in 2012, almost 68% of its stock market was concentrated in banking and financial services. Thirdly, the analysis and argument carried forward in our study also establish an important basic understanding of the relationship between single and regional stock market characteristics of emerging equity markets, particularly the GCC markets, and the implications of these features on the status of

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

243

the efficiency in these markets. Discounting the impact of thin trading because it has become obsolete, the two identified stock market features, including weak foreign participation and high concentration in banking businesses, may continue to have a negative influence on the efficiency of GCC markets. Lastly, although the results of our study complement the empirical results of Al-Ajmi and Kim (2012) and Bley (2011) who applied a similar range of robust parametric and nonparametric tests to prove the inefficient status of GCC stock markets, we went the extra mile by being the first GCC study to employ the technique of cointegration analysis to examine the collective weak-form efficiency of GCC stock markets.

8. Implications of the results The implications of the results of our study are of importance to the academic body of knowledge, and to investors and policy makers in the GCC countries. For academics, the use of cointegration testing applications in order to show the presence of a long-term relationship of GCC stock market prices provides supporting evidence that GCC stock markets are inefficient as a regional stock market. To the best of our knowledge, none of the previous studies have used the application of the cointegration analysis of stock market prices to examine the collective weak-form efficiency in the GCC stock markets. Therefore, using the application of the cointegration testing to examine the efficiency of stock markets may extend the horizon of the literature of GCC stock market efficiency, in which a long-term relationship of stock prices can be considered as a tool for testing the weak-form efficiency of a group of stock markets in the Middle East, and elsewhere. For investors, the conclusion drawn about the rejection of the weak-form efficiency of the GCC stock returns implies that future successive price changes can be predicted from past price changes, as not all past stock information is fully incorporated in current prices. Hence, investors may apply, for example, a “beat the market” strategy against a “hold the market” strategy to produce excessive returns. GCC stock markets provide valuable investment opportunities for regional and global investors who seek diversification opportunities in markets that have simultaneously positive mean returns and predictable price movements, with the exception of the Bahraini stock market that has negative mean returns. The documented presence of a long-term relationship among GCC stock prices implies that investors in an ‘X’ GCC stock market may be able to predict the movement of ‘Y’ GCC stock market from past movements of the former, due to the presence of a long-term price bond between stock markets X and Y. For GCC policy makers, the results of our study are an alarming outcome that may hinder longterm growth plans set forth by GCC policy makers, that is, to reduce reliance on the non-renewable oil revenues in order to promote local and regional economic growth. GCC leaders’ plans to expand the size of their markets along with enhancing the market efficiency may be at risk, as GCC stock markets share common persistent stock market characteristics that would disable the presence of such efficiency for GCC stock markets. We have identified two persistent stock market features that GCC markets share. First, there is weak foreign participation caused by the presence of signals of information asymmetry, in turn caused by the presence of foreign ownership restriction and weak financial market development. The financial market development proxies provided, including measures of ease of doing business, transparency of public and private market sectors, and adequacy of digital financial disclosure, indicate that the GCC financial market is informationally inefficient. Second, the high industry concentration in crisis-prone industries such as the banking business may negatively influence GCC markets’ efficiency status, for just as while banking and financial linkages promote the transfer of offshore financial crises into GCC markets, in turn a financial crisis causes a rapid establishment of increasing frequencies of serial correlations among current and past price changes. The conclusions drawn by this study about the rejection of the weak-form efficiency of the GCC stock markets imply the presence of asymmetric information among GCC investors, by which not all investors have the same access to all publicly available information. Thus, the lack of stock market efficiency among GCC markets is likely to be a disincentive, and may work as a negative sentiment signal for serious regional, migrant, and offshore investment capital to invest in GCC markets, consequently leading to continuing reliance on the nonrenewable oil resources to boost local and regional economic growth.

244

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

9. Conclusion With the growing interest of investors in identifying investment opportunities, of academics in understanding the working mechanism of equity markets, and of policy makers in building positive market sentiment, efficiency of developing stock markets matters. Efficiency of stock markets provides a theoretical and predictive model imperative for the operation of equity markets that would assist local, regional, and global investors to identify mispriced assets, leading to enhancing their riskadjusted returns. Ensuring market efficiency is also vital to regulators of equity markets, in the sense that efficiency of stock markets works as a safeguard mechanism that protects markets from distortions resulting from the presence of information asymmetry amongst market participants, which in turn acts as a disincentive to serious regional and global investment capital. We employed daily index prices denominated in local currencies from the end of December 2003 to the end of January 2013 using a number of parametric and nonparametric, unit root, and cointegration tests. We aimed to answer an important research question, whether GCC stock markets are weak-form efficient as single stock markets or as a regional stock market. The results of our study showed that all GCC stock markets are not weak-form efficient, in the sense that current stock price movement can be predicted from past price movements. When we examined the weak-form efficiency of GCC stock markets from a collective perspective, we found that GCC stock market prices were cointegrated in the long run, meaning that the movement of individual GCC stock market prices can be predicted by the past movements of other GCC market prices. The implications of our results imply that investors can easily identify mispriced GCC stocks by observing past price changes within an individual GCC stock market. Investors who wish to spot mispricing in one GCC market by looking at historical price changes of another GCC market can do so, as GCC stock market index prices are cointegrated in the long term. Institutional investors such as fund managers who seek to diversify their investment portfolios by considering GCC stock markets as an alternative channel of investment can take advantage of the findings of our study. The implications of our findings are also of importance for those academics who wish to understand the working mechanisms and the efficiency characteristics of these regional stock markets. The outcomes of our study are also of great importance to GCC policy regulators. Better informed about the current efficiency status of their equity markets, they can improve the flow of information among market participants in order to increase the attractiveness of their equity markets to regional and international investment capital, leading to enhanced market sentiment and to a greater role for their equity markets in boosting local and regional economic growth. As our results established the linkage between the composition of emerging equity markets on the status of their equity market efficiency, future research endeavours may consider an empirical examination of this linkage. Acknowledgements We thank (the Editor) and the anonymous referee for their comments and suggestions. We also thank Associate Professor Robert Bianchi for his initial comments on initial draft. Thanks also to Dr. Philip Robertson for his proofreading assistance. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.ribaf.2014.09.001. References Abdmoulah, W., 2010. Testing the evolving efficiency of Arab stock markets. Int. Rev. Financ. Anal. 19 (1), 25–34. Abraham, A., Seyyed, F., Alsakran, S., 2002. Testing the random walk behavior and efficiency of the Gulf stock markets. Financ. Rev. 37 (3), 469–480. Abraham, A., Seyyed, F.J., 2006. Information transmission between the Gulf equity markets of Saudi Arabia and Bahrain. Res. Int. Bus. Financ. 20 (3), 276–285.

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

245

Al-Ajmi, J., Kim, J.H., 2012. Are Gulf stock markets efficient? Evidence from new multiple variance ratio tests. Appl. Econ. 44 (14), 1737–1747. Al-Khazali, O., Darrat, A.F., Saad, M., 2006. Intra-regional integration of the GCC stock markets: the role of market liberalization. Appl. Financ. Econ. 16 (17), 1265–1272. Al-Khazali, O.M., Ding, D.K., Pyun, C.S., 2007. A new variance ratio test of random walk in emerging markets: a revisit. Financ. Rev. 42 (2), 303–317. Al Ashikh, A.I., 2012. Testing the weak-form of efficient market hypothesis and the day-of-the-week effect in Saudi stock exchange: linear approach. Int. Rev. Bus. Res. Pap. 8 (6). Al Janabi, M.A., Hatemi-J, A., Irandoust, M., 2010. An empirical investigation of the informational efficiency of the GCC equity markets: evidence from bootstrap simulation. Int. Rev. Financ. Anal. 19 (1), 47–54. AlKhazali, O., 2011. Does infrequent trading make a difference on stock market efficiency?: Evidence from the Gulf Cooperation Council (GCC) countries. Stud. Econ. Financ. 28 (2), 96–110. Ariss, R.T., Rezvanian, R., Mehdian, S.M., 2011. Calendar anomalies in the Gulf Cooperation Council stock markets. Emerg. Mark. Rev. 12 (3), 293–307. Azad, A., 2009. Random walk and efficiency tests in the Asia-Pacific foreign exchange markets: evidence from the post-Asian currency crisis data. Res. Int. Bus. Financ. 23 (3), 322–338. Azad, A., Azmat, S., Fang, V., Edirisuriya, P., 2014. Unchecked manipulations, price–volume relationship and market efficiency: evidence from emerging markets. Res. Int. Bus. Financ. 30, 51–71. Bekaert, G., Harvey, C.R., Ng, A., 2005. Market integration and contagion. J. Bus. 78 (1), 39–69. Bhattacharyya, S.C., 2011. Impact of high energy prices. Energ. Econ., 441–462. Bley, J., 2011. Are GCC stock markets predictable? Emerg. Mark. Rev. 12 (3), 217–237. Boehmer, E., Wu, J.J., 2013. Short selling and the price discovery process. Rev. Financ. Stud. 26 (2), 287–322. Boughanmi, H., 2008. The trade potential of the Arab Gulf Cooperation Countries (GCC): a gravity model approach. J. Econ. Integr. 23 (1), 42–56. Butler, K.C., Malaikah, S.J., 1992. Efficiency and inefficiency in thinly traded stock markets: Kuwait and Saudi Arabia. J. Bank. Financ. 16 (1), 197–210. Chiang, S.-M., Lee, Y.-H., Su, H.-M., Tzou, Y.-P., 2010. Efficiency tests of foreign exchange markets for four Asian countries. Res. Int. Bus. Financ. 24 (3), 284–294. Choi, J.J., Lam, K.C., Sami, H., Zhou, H., 2013. Foreign ownership and information asymmetry. Asia-Pacif. J. Financ. Stud. 42 (2), 141–166. Chow, K.V., Denning, K.C., 1993. A simple multiple variance ratio test. J. Econometr. 58 (3), 385–401. Dahel, R., Laabas, B., 1999. The behavior of stock prices in the GCC markets. In: Paper presented at the Economic Research Forum Working Papers. Elango, D., Hussein, M.I., 2008. An empirical analysis on the weak-form efficiency of the GCC markets applying selected statistical tests. Int. Rev. Bus. Res. Pap. 4 (1). Fifield, S.G., Jetty, J., 2008. Further evidence on the efficiency of the Chinese stock markets: a note. Res. Int. Bus. Financ. 22 (3), 351–361. Gandhi, D.K., Saunders, A., Woodward, R.S., 1980. Thin capital markets: a case study of the Kuwaiti stock market. Appl. Econ. 12 (3), 341–349. Granger, C.W., 1986. Developments in the study of cointegrated economic variables. Oxford Bull. Econ. Stat. 48 (3), 213–228. Guidi, F., Gupta, R., 2013. Market efficiency in the ASEAN region: evidence from multivariate and cointegration tests. Appl. Financ. Econ. 23 (4), 265–274. Guyot, A., Lagoarde-Segot, T., Neaime, S., 2014. Foreign shocks and the international cost of equity. Evidence from the MENA region. Emerg. Mark. Rev. 18, 101–122. Hammoudeh, S., Choi, K., 2006. Behavior of GCC stock markets and impacts of US oil and financial markets. Res. Int. Bus. Financ. 20 (1), 22–44. Harvey, C.R., 1995. Predictable risk and returns in emerging markets. Rev. Financ. Stud. 8 (3), 773–816. Healy, P.M., Palepu, K.G., 2001. Information asymmetry, corporate disclosure, and the capital markets: a review of the empirical disclosure literature. J. Acc. Econ. 31 (1), 405–440. Hussainey, K., Al-Nodel, A., 2009. Corporate governance online reporting by Saudi listed companies. Res. Acc. Emerg. Econ. 8, 39–64. Ismail, T. H. (2002). An empirical investigation of factors influencing voluntary disclosure of financial information on the internet in the GCC countries. Available at SSRN 420700. Jensen, M.C., 1978. Some anomalous evidence regarding market efficiency. J. Financ. Econ. 6 (2), 95–101. Jiang, L., Kim, J.B., 2004. Foreign equity ownership and information asymmetry: evidence from Japan. J. Int. Financ. Manage. Account. 15 (3), 185–211. Joshi, P., Al-Modhahki, J., 2003. Financial reporting on the Internet: empirical evidence from Bahrain and Kuwait. Asian Rev. Account. 11 (1), 88–101. Kim, J.H., Shamsuddin, A., 2008. Are Asian stock markets efficient? Evidence from new multiple variance ratio tests. J. Empir. Financ. 15 (3), 518–532. Klapper, L.F., Love, I., 2004. Corporate governance, investor protection, and performance in emerging markets. J. Corporate Financ. 10 (5), 703–728. Lagoarde-Segot, T., Lucey, B.M., 2006. Financial vulnerability in emerging markets. Evidence from the Middle East and North Africa. The Institute for International Integration Studies Discussion Paper Series, iiisdp114. Lee, C.-C., Lee, J.-D., Lee, C.-C., 2010. Stock prices and the efficient market hypothesis: evidence from a panel stationary test with structural breaks. Jpn. World Econ. 22 (1), 49–58. Lence, S., Falk, B., 2005. Cointegration, market integration, and market efficiency. J. Int. Money Financ. 24 (6), 873–890. Lim, K.-P., Brooks, R.D., Kim, J.H., 2008. Financial crisis and stock market efficiency: empirical evidence from Asian countries. Int. Rev. Financ. Anal. 17 (3), 571–591.

246

F. Jamaani, E. Roca / Research in International Business and Finance 33 (2015) 221–246

Lo, A.W., MacKinlay, A.C., 1988. Stock market prices do not follow random walks: evidence from a simple specification test. Rev. Financ. Stud. 1 (1), 41–66. Malkiel, B.G., Fama, E.F., 1970. Efficient capital markets: a review of theory and empirical work. J. Financ. 25 (2), 383–417. Mobarek, A., Fiorante, A., 2014. The prospects of BRIC countries: testing weak-form market efficiency. Res. Int. Bus. Financ. 30, 217–232. Moustafa, M.A., 2004. Testing the weak-form efficiency of the United Arab Emirates stock market. Int. J. Bus. 9 (3), 309–325. Neaime, S., 2002. Liberalization and financial integration of MENA stock markets. In: Paper presented at the A paper presented at the ERF’s 9th annual conference on Finance and Banking. Neaime, S., 2005. Financial market integration and macroeconomic volatility in the MENA region: an empirical investigation 1. Rev. Middle East Econ. Financ. 3 (3), 231–255. Neaime, S., 2012. The global financial crisis, financial linkages and correlations in returns and volatilities in emerging MENA stock markets. Emerg. Mark. Rev. 13 (3), 268–282. Onour, I.A., 2004. Testing weak-form efficiency of Saudi stock exchange market (Available at SSRN 611209). Onour, I.A., 2009. Testing efficiency performance of Saudi stock market. J. King Abdulaziz Univ.: Econ. Adm. 23 (2), 15–27. Peress, J., 2010. Product market competition, insider trading, and stock market efficiency. J. Financ. 65 (1), 1–43. Phillips, P.C., Perron, P., 1988. Testing for a unit root in time series regression. Biometrika 75 (2), 335–346. Rao, D., Shankaraiah, K., 2003. Stock market efficiency and strategies for developing GCC financial markets: a case study of the Bahrain stock market. Arab Bank Rev. 5 (2), 16–21. Risso, W.A., 2008. The informational efficiency and the financial crashes. Res. Int. Bus. Financ. 22 (3), 396–408. Sandoval Junior, L., Franca, I.D.P., 2012. Correlation of financial markets in times of crisis. Phys. A: Stat. Mech. Appl. 391 (1), 187–208. Squalli, J., 2006. A non-parametric assessment of weak-form efficiency in the UAE financial markets. Appl. Financ. Econ. 16 (18), 1365–1373. World Economic Forum, 2014. The Global Competitiveness Report, Available from: http://www.weforum.org/reports? filter[type]=Competitiveness (accessed 10.05.14). Wright, J.H., 2000. Alternative variance-ratio tests using ranks and signs. J. Bus. Econ. Stat. 18 (1), 1–9.