Insert Table 11 about here
The study fills the gap in limited research focusing on Africanstock market integration. The main purpose of this study was to empirically determine stock market integration among Africanstock market and subsequently examine the long run relationship between the SouthAfricanstock market and selectedAfricanstockmarkets. The results suggest that cointegration among these Africanstockmarkets is incomplete hence the stockmarkets are not fully integrated. These results concur with finding by Dahalan (2012) who found a weak cointegrating relationship among Africanstockmarkets. The study incorporated the regime switching VECM to explain the long run relationship between the SouthAfricanstock market and selectedAfricanstockmarkets. The estimated parameters have varying effects across regimes. In some instances, parameters have reversed signs, or the influence is either significant or insignificant across regimes. The regime switching VECM are compared using the log likelihood and the information criterion. Results suggest that the regime switching VECM better fits the data well and performs better than the linear VECM.
Recent developments in the financial and capital markets have led to ongoing research to provide insight into financial integration across global markets. Whilst financial needs vary with different markets, capital markets respond to satisfy these varying needs both at domestic and international levels, in developing, emerging and developed market-based economies. According to financial theory, the more integrated markets are, the more efficient they are compared to fragmented financial markets. According to Chattopadhayay (2006), stock market integration is of vital importance as it contributes to both static and dynamic gains. Static gains include achieving economies of scale, allocative efficiency through trade creation as well as spillover effects derived from market expansion. Dynamic gains result from the elimination of trade barriers and trade diversion between countries. The markets become integrated if there is no arbitrage and the law of one price exists such that all assets bearing similar risk command the similar expected return in different markets (Bekaert, 1995).
The few studies that have examined first moment linkages among Africanstockmarkets with their regional and/or global counterparts include those by Lamba and Otchere (2001), Piesse and Hearn (2002), Collins and Biekpe (2003a), Alhassan (2006) and Chinzara and Aziakpono (2009a). Lamba and Otchere (2001) provide the first comprehensive analysis of dynamic interactions of seven African equity markets with their regional (African) and global counterparts using a multivariate VAR model between 1988 and 2000. The results indicate integration along regional lines, especially among South Africa, Namibia and Zimbabwe. Furthermore, with the exception of Namibia and South Africa, there is little evidence of interdependence of Africanmarkets with their global counterparts. Similar results for Namibia and South Africa are obtained by Piesse and Hearn (2002) who conduct cointegration tests on the three dominant Southern African Customs Union (SACU) member states, namely South Africa, Botswana and Namibia for the period 1990 to 2000. Collins and Biekpe (2003a) use Granger causality tests to analyse the interdependence among returns of eight Africanmarkets and an adjusted correlation coefficient (as in Forbes and Rigobon, 2002) to analyse the extent to which these countries were affected by the 1997 Asian crisis. The Granger causality tests reveal linkages among regional lines, specifically for South Africa and Zimbabwe. With the exception of the two largest Africanmarkets, Egypt and South Africa, the Forbes and Rigobon (2002) adjusted correlation coefficients suggest no evidence of contagion from the Asian crisis.
There are studies which have employed the GARCH model utilising daily data for the period 1994-2012. These studies include Fratzscher (2001), Batram, Taylor and Wang (2006), Wang and Moore (2008), Yoshida (2009), and Baumohl (2013). The majority of these studies concur with those conducted using weekly data. For instance, Berben and Jansen (2005) utilised weekly data for the period 1980-2003 and concluded that stock market integration advanced in the late 1980s and 1990s in Europe; that is, the stockmarkets from the developed nations prove to be integrated. More so, monthly studies provide the same results for different markets. Buttner and Hayo (2008), Johansson (2009), Guesmi (2011), Arouri (2012), and Berger and Pozzi (2012) conducted the studies within the range of 1970-2011. All the studies show that integration is increasing for the different stockmarkets. Other studies (Ntim, 2012; Onour, 2009) utilised daily data for the different African countries to determine why Africanstockmarkets should formally harmonise and to determine financial integration of North Africanstockmarkets, utilising the parametric and non-parametric variance ratios and Breitung (2001) rank test. The results from the study conducted by Ntim (2012) suggest that there is a weak form of information efficiency, and the author concluded that formal harmonisation may improve information efficiency. Also, results from Onour (2009) provided evidence of multivariate and bivariate non-linear cointegration between the three North Africanstockmarkets (Egypt, Morocco and Tunisia).
Ghosh et. al. (1998) checked the individual integration of nine Asia pacific markets with either the US or Japanese stockmarkets, but they did not find any evidence that US or Japanese stock market movements dominates these markets. Phylaktis (1999) also examined Asia Pacific Basin countries to investigate whether the Japan has play a more influencing role in this region than USA. Using Impulse-response analysis for speed of adjustments and long-run comovements of real interest rates, they found that these countries are closely linked with world financial markets. Their relationship is stronger with Japan than with USA. Haung et al (2000) explored the causality and cointegration relationship among the US, Japan and several South East Asian countries including recently established markets in China; Shenzhen and Shanghai exchanges. They found no cointegration among these markets except between Shanghai and Shenzhen.
Botswana; Ghana; Kenya; Namibia; Nigeria; Tanzania; Zambia) for the January 2002 – July 2015 period. Given the research questions our work addresses (and which were outlined earlier in this section), our results can be summarized as follows. Investors herd in all eight markets, something that can be ascribed to the low transparency levels prevailing in frontier stock exchanges that reduce the quality of their informational environment, leading investors to resort to herding as a means of inferring information by tracking their peers’ trades. Smaller stocks amplify the magnitude of herding, since the latter grows larger for equal- (compared to value-) weighted estimations, something hardly surprising, given the greater informational uncertainty surrounding smaller stocks that prompts investors to herd more when trading them. Herding is not found to exhibit significant asymmetries conditional on market returns, as it appears significant irrespective of the market’s directional movement in most cases. On the other hand, herding appears significant (or stronger, compared to high volatility days) mainly during days of low volatility, with this asymmetric pattern, however, growing weak when partitioning our sample period to account for the 2007-2009 global financial crisis. Although “domestically” motivated herding is significant across all eight markets, the same cannot be argued for herding induced by the US and SouthAfrican market returns, the presence of which is confirmed on only a small number of occasions; similarly, the return dynamics of a regional economic initiative’s member-markets are found to motivate herding in each other very rarely. These results are very interesting, as they indicate that investors’ behaviour in African frontier markets is not significantly affected by non-domestic factors and are in line with extant research denoting the overall low levels of integration of frontier markets within the global financial system.
In order to confirm the usability of the data for further analysis in the study, additional tests are made to identify whether or not the data applied in the study contain unit root and is stationery. This is done using the Augmented Dickey Fuller (ADF) test and Kwiatkowski-Phillips-Schmidt-Shin (KPSS) respectively. The ADF is known to have a null hypothesis of the presence of unit root, while the alternative hypothesis is that there is no unit root. In the ADF test, the negative number under t-statistic is looked at and the more negative it is, the stronger the rejection of the null hypothesis. From the results below, all the variables are statistically significant at 1% level indicating that the null hypothesis is rejected. Stationarity of the series is however determined with the aid of the KPSS test. The null hypothesis is that there is stationarity of the data. From the test results, there was insignificance of probability at 1% significance level, which means the null hypothesis is accepted which makes the data valid for further analysis. Furthermore, based on the selected model for the study, there is the need to make further tests to check for the ARCH tests. In doing so, firstly, the Lagrange multiplier test which is made to assess ARCH effects in the series is conducted for autoregressive conditional heteroskedasticity on residuals. The obtained Obs*R-squared which are large enough are also accompanied by significant Prob Chi-square at 1% significance level. This is an indication that the data has ARCH effects and confirms the model selection for the study. Also, the heteroskedasticity test is carried on the residuals at selected lags (15). The Ljung-Box Q- test which is used here is a way to test for autocorrelation using joint multiple lags. From the table below, the LB 15 signifies
However, negative consequences of increased global financial market integration are, inter alia, the increasingly undesirable return and volatility spillover effects across stockmarkets of different countries. This is particularly the case during periods of financial crises (Sugimoto et al., 2014). Increasing global financial market integration reduces the insulation of the domestic market against global structural shocks, and exposes the domestic market to the risk of financial market contagion (Choudry & Jayasekera, 2014). Notably, the 2007 United States’ (US) sub -prime housing market financial crisis expanded to the EU as a sovereign debt crisis in 2008, which ultimately led to declines in stock market prices of both advanced economies and EMEs, i.e. the period widely referred to as the 2007-2009 global financial crisis. It is important to note that literature abounds regarding the transmission of stock market returns and volatility spillover between advanced economies and EMEs (Sugimoto et al., 2014; Piesse & Hearn, 2005; Balli et al., 2015; Duncan & Kabundi, 2013; Öztürk & Volkan, 2015; Korkmaz, Çevik & Atukeren, 2012).
Subsequent research was done comparing Africanmarkets and the main conclusion was that Africanmarkets were not well cointegrated with each other, with the exception of a few cases where Namibia was found to be cointegrated with South Africa. The Africanmarkets were also found to be poorly cointegrated with global markets when compared pair-wise, but certain combinations of portfolios from various markets revealed the presence of cointegration (Appiah-Kusi and Pescetto, 1998; Piesse and Hearn, 2002; Wang et al., 2003; Collins and Abrahamson, 2004). As Piesse and Hearn note, since the late 1980s, the IMF and World Bank have sponsored structural adjustment programs in the equity markets of most African countries. This, along with more easily available stock price information, has increased foreign participation in Africanmarkets. Africanmarkets have traditionally not been regarded as safe by foreign investors due to concerns about political stability, corruption, lack of proper legal and commercial structure amongst other things. However, the structural adjustment of the financial sector and the increased political integration in the form of trade unions have increased the chances of cointegration amongst Africanmarkets and Africanmarkets with the rest of the world (Piesse and Hearn, 2005: 38). Bekaert and Harvey (2002) also conducted research and summarized key findings that have been made over the last few decades and found that traditional models do not fit the data as well as with developed markets. In addition, the data that are available for emerging markets are not as detailed and extensive as that available for developed markets.
Furthermore, both the MOW and DOW effects are combined in some studies. For instance, Lei and Gerhard (2005) investigated calendar effects in the Chinese stock market, especially monthly and daily effects. Returns of the market index in Shanghai and Shenzhen stock exchanges were used to analyse the monthly and daily effects in stock returns. Results revealed that the highest returns could be achieved after the Chinese year-end in February while Mondays are seen to be weak and Fridays showed signiﬁcant positive average returns. Yet the daily effect has a minor magnitude and relevance for determining average returns compared to monthly effects. Similarly, Rossi (2007) examined the calendar anomalies in stock returns in South America from 1997 to 2006, focusing on the existence of DOW effects and the monthly patterns in Argentina, Brazil, Chile and Mexico. In full period, it was concluded that there existed the traditional positive Friday effect in Brazil and in Chile; the returns had been lowest on Mondays. In addition, the study documented positive returns on Wednesdays and Fridays. In Mexico highest returns appeared on Wednesdays. For Argentina, there was no record of DOW anomaly. These results change when examined over two sub-periods. Overall, there is absence of monthly anomalies in full period and first sub-period, but January effect is found in Argentina in second sub-period. Additionally, Lukas (2009) studied stock market seasonality with focus on DOW effect and January effect by analysing 30 stocks traded on the German Stock Exchange from 1995 to 2009. By adopting a dummy variable approach to investigate Monday effect and the September effect, it was confirmed that the DOW effect started disappearing in the second half of 1990s. Moreover, Martin (2011) carried out a comprehensive review of the literature on calendar anomalies from 1915 to 2009. It was found that intraday, holiday and intra month effects still exist, the weekend effect seems to have disappeared and the January effect has halved.
Conclusions from this paper are relevant for portfo- lio diversification strategies and policy makers. Firstly, this strong integration specific at larges scales means that Africanstockmarkets excepted South Africa do not re- act immediately to world financial shocks. This long run integration could be an opportunity of capital diversifica- tion for financial agents in small time scales. Secondly this large scales integration could be explained by chan- nels of transmission of the financial information flow be- tween the African and external stockmarkets that are not sufficiently developed (non-efficient financial institu- tions and a low representation of African or world enter- prises listed in Africanstockmarkets. . . ). The authorities concerned should therefore redouble their efforts in view of the substantial economic advantages stemming from devel- oped stockmarkets(Hicks, 1969; Levine, 1997; Calder´on and Kubota, 2009).
question of his effective relationship with the outside. there are some studies. Collins and Biekpe (2003a) suggest that the most integrated markets in Africa, which are Egypt and South Africa are suffered from contagion during Hong Kong’s 1997 crisis. Collins and Biekpe (2003b) find that the interdependence of African financial markets in regional blocks are falling and excepted South Africa and Egypt, there is no integration with the global emerging markets. Wang et al. (2003) using co-integration and error correc- tion model show that Africanstockmarkets have an inte- gration which varies in time and that appears to decrease after the Asian crisis of 1997. Adjasi and Biekpe (2006) find a long-run unique relationships betweenAfricanstockmarkets and also short-run dynamic returns from others Africanstockmarkets that affect SouthAfrican and Ghana- ian stock market. Agyei-Ampomah (2008) uses the mea- surement’s method of the integration score market proposed by Barari (2004) and find low correlation levels betweenAfricanmarkets themselves and with global stockmarkets. More recently Boamah (2013) through a multi-factor pric- ing model, indicates that the integration of African finan- cial markets evolve through time. He underlines that this evolution may be the result of global economic conditions changing 6 . Sugimoto et al. (2014) using the 2012 Diebold
The results of the wavelet-based unit root tests are presented in Table 1 below. From the results, it can be deduced that the null hypothesis for all the African countries studied are rejected using the three methods employed. This suggests that when frequency domain is factored into stock market efficiency framework, no evidence abound to support the existence of random walk hypothesis for the studied African economies. The implication is that the selectedmarkets are inefficient, and are capable of providing the platform to support economic growth (Pele and Vioneagu, 2008; Narayan et al, 2016; Westerlund and Narayan, 2015; Narayan et al, 2015). The results also suggest that arbitrage opportunities exist in the studied markets, for instance, if stock prices are mean reverting, it is possible for investors to make abnormal profit by predicting price movements based on historical data using technical analysis. This further suggests that shocks to assets prices due to global financial crisis will only result in temporary deviation from long run growth path and that stock prices will return to the long run equilibrium, thus foreign investors are encouraged to maximize the arbitrage opportunities in these markets.
trading days). The name of the country whose stock market maximizes the multiple correlation against a linear combination of the rest of variables is in upper-left. This stock exchange can be a potential leader or follower for the others stockmarkets. In our case, across all scales, the EGX30 (Egypt) is a potential leader or follower except at scale 1 where it is the TOP40 (South Africa). The results from the wavelet multiple correlation are confirmed by the wavelet multiple cross-correlation. For scale 7, we note an asymmetry (negative-skewness) which means that on this scale, the EGX30 lags the others indices. Compared to other studies of stockmarketsintegration using the same methods in Europe and Asia (Tiwari et al., 2013; Fern´andez-Macho, 2012), Africanstockmarkets are far from integrated. For the construction of the confidence intervals, we used the estimators proposed by Whitcher et al. (2000). They are robust to the non-normality distribution.
This raises the important question as to how Africanmarkets are progressing with regards to market efficiency. There is therefore a constant need for current analysis of the markets to understand the movements and patterns that investors face. Although crucial, the generally recommended policy implications for investors imply that patterns are not subject to changes nor that shares are traded only on an extremely long-term basis. In contrast to previous literature (see Enowbi et al., 2009; Tachiwou, 2010; Mbululu & Chipeta, 2012), this paper proposes that the day-of-the-week effects are subject to changes and movement and tests for differences in the day-of-the-week effect across time for South Africa, Botswana, Nigeria, Morocco, and Zambia. Further, the previous literature utilises Arch, Garch, and Regime Switching models (see Basher & Sardosky, 2006; Yalcin & Yucel, 2006; Alagidede, 2008), which only consider the mean and standard deviation of returns, thereby leaving out higher statistical moments in the search for seasonal patterns in returns. This paper performs a direct test on skewness and kurtosis, which better captures the distributional asymmetries of daily returns.
The subsequent parts of this research are structured as follows. Chapter 2 discusses relevant theories relating to this research, which include the Random Walk Theory of Regnault (1863), the Efficient Market Hypothesis (EMH) of Fama (1965, 1970), Capital Asset Pricing Model (CAPM) of Sharpe (1964), Prospect Theory of Kahneman and Tversky (1979, 1992) and Overreaction Hypothesis of DeBondt and Thaler (1985; 1987). Chapter 3 presents an overview of prior literature on overreaction hypothesis in international and SouthAfricanstockmarkets. The review explores the empirical evidence of investor overreaction as well as alternative explanations suggested by various researchers. Chapter 4 describes the process of data collection, the sample, the methodology used as well as the motivation behind the methodology employed. Chapter 5 presents and interprets the post-formation results of the winner and loser portfolios over short-term and long-term formation periods while Chapter 6 undertakes to investigate the specific timing of mean reversion for portfolios constructed under different formation periods with reference to the various stages in the SouthAfrican economic cycle. Chapter 7 presents summary and conclusion on findings.
In a similar study, Voronkova (2004) examines the long-run links between the three emerging CE markets (the Czech Republic, Hungary, and Poland), three developed European stockmarkets (Great Britain, France, and Germany), and the US. She uses weekly data that covers a period of almost 10 years, from 1993 to 2002. Voronkoava applies Engle and Granger and Johansen bivariate and multivariate tests and compares them to the findings of the Gregory – Hansen test. She uses this approach in order to investigate whether the Gregory and Hansen methodology could possibly provide more eveidence on the presence of long-run relationships that the conventional cointegrations tests would not detect. The results point towards the existence of six additional cointegration relationships (one within the group of Central European markets and five between the Central European and the mature markets). Most importantly, Voronkova finds evidence of links between the emerging CE markets within the region and globally that is stronger than has previously been reported. Unlike the previous study of Gilmore and McManus (2002) her study supported the hypothesis that the emerging CE markets have become increasingly integrated with the world markets.
Integration among stockmarkets has received considerable attention by policy makers as well as finance specialists. There are a number of reasons to support this move. Firstly, it can be argued that it provides opportunities in risk sharing among integrated markets (Marashdeh & Shrestha, 2010). In addition, it contributes to financial stability by enhancing competition and efficiency in allocation of resources, as well as reducing the cost of capital and price volatility among integrated markets (Tai, 2007). Integration of stockmarkets also plays an important role in promoting domestic savings, investment and could positively affect total factor productivity and economic growth (Levine, 2001). However, there are a number of studies which does not support stock market integration. Most notably those that argue that stock market integration could pose a major risk of contagion as was evidenced during the Asian crisis of 1997 and the global financial crisis.
In the local context, themost accurate source of industrial action-related data is the Department of Labour’s annual Industrial Action Report. The 2013 edition summarises the past four years and notes the increase of industrial action, citing 51 incidents in 2009, 74 in 2010, 67 in 2011, 99 in 2012 and 114 recorded in 2013.It notes an increase in wages lost in 2013 amounting to 6.7 billion rand, up from 6.6 billion rand in 2012, with the average wage increase at 8%, a level above inflation. Added to this, the SouthAfrican Transport and Allied workers Union (SATAWU) and National Union of Mineworkers (NUM) had the biggest participation strike rate of members and, due to sheer volume size, were able to negotiate a 25.7% and 17.4% wage increase respectively. The Departmentelaborates on this by stating that low wages, rising inequality and tough economic conditions had led analysts to predict toughnegotiations in 2013. This was also influenced by above-average wage increases granted in 2012. It was found that the mining, manufacturing and community industries were hardest hit, case in point being the manufacturing industry, which contracted 6.6%, forcing car giant producer BMW to stop its expansion projects in South Africa and look for new manufacturing locations overseas. There’s a 25% unemployment rate in the mining, manufacturing and community sectors, which is set to continue to negatively affect the economy in the next two to three years.
Stock index futures contracts can be used to hedge the risk. Hedging uses futures markets to reduce risk of a cash (spot) market position. According to Hull (2000, p. 66), when the relationship between the cash price and the price of a futures contract is very close, the hedge is more effective. However, because this relationship is usually not perfect (spot and futures positions do not move together), the hedge is a cross-hedge. In this case, the hedger should trade the right number of futures contract to control the risk. In other words, the determination of the optimal hedge ratio 1 (minimum variance hedge ratio, or MVHR) is required. MVHR is the optimal amount of futures bought or sold expressed as a proportion of the cash position. It is important for the hedger to be able to identify the number of contracts needed to hedge the portfolio. Thus, the hedge ratio (HR) will be used, so one can choose the right number of futures contracts minimising risk. The HR is the number of futures contracts bought, or sold, divided by the number of spot contracts whose risk is being hedged. Several measures have been proposed for the HR computation. Usually, the HR is estimated from an OLS regression of cash on futures prices. The method is introduced by Ederington (1979), and Anderson and Danthine (1980). The slope coefficient of the OLS regression is the MVHR, which is constant over time. An alternative estimation of the optimal HR is based on the phenomenon that cash and futures prices display volatility clustering, and, hence, GARCH models are to be