Returns Have Focused On Developed Countries

Published Date: 02 Nov 2017

1.0 Introduction

The stock market is the barometer of a country’s economy. At any point in time the stock market reflects how well the economy is performing. A well-functioning stock market promotes development and growth of a country. The stock market encourages household to save and invest in financial assets. As a result an avenue for capital is possible for firms which need to meet rapid expansion and finance investment projects. Thus the stock market facilitates transfer of funds from surplus to deficit units and serves as a tool for the mobilization of resources in an economy. In this retrospect, the stock market is the focus of investors and policymakers due to the benefits that the economy perceives with it.

1.1 Motivation for Research

Most of the empirical studies conducted to examine the effect of macroeconomic forces on stock market returns have focused on developed countries. With the liberalization of emerging economies and their removal of foreign capital controls, capital flows has risen continuously to those markets. However there is a lack of research on emerging economies in spite of rising capital inflows to those markets. The structure of emerging markets is different from that of matured markets. As a result stock market index movement may vary. Therefore it is of paramount importance to see how these emerging markets react to innovations in fundamental macroeconomic variables.

Hence to overcome the gap in the literature this study will examine the impact of macroeconomic forces in Mauritius which is an emerging economy. Furthermore policymakers can have some insights regarding policy implementation to control volatility in macroeconomic variables because this would affect the global return of investors which in turn will affect the level of consumption and capital inflows in the market. Since stock prices reflect profitability and profitability itself is related to economic activity, fluctuations in stock prices can lead the economic direction. Therefore this study can also be helpful to policymakers for strategy formulation and implementation to enhance the efficiency of the Mauritian stock market.

1.2 Objective of the Study

The aim of the study is to investigate how macroeconomic variables affect stock market returns in Mauritius. The selected macroeconomic variables are inflation (LCPI), exchange rate of rupee vis-à-vis the US dollar (LEXC), money supply (LM2), overall yield on treasury bills (LYIELD) and oil prices (LOIL). The causality between these variables will also be examined.

1.3 Data and Methodology

The study will employ the Johansen and Jesilius multivariate cointegration test to investigate the long-run relationship between the stock market and the macroeconomic variables. The Granger-causality test is used to examine whether there is a causal relationship between the returns of SEM and the macroeconomic fundamentals. The data to be used in the study will be obtained from the Bank of Mauritius and the Stock Exchange of Mauritius.

1.4 Organisation of the Study

The dissertation is structured as follows. Chapter 1 introduces the topic; Chapter 2 will present a review of the theoretical and empirical literature. An overview of the development of the Stock Exchange of Mauritius since its existence will be provided in Chapter 3. Chapter 4 embodies the research methodology. The empirical model and data required will be specified. Chapter 5 presents the empirical analysis and interpretation of results. Chapter 6 highlights the conclusion and policy recommendations.

CHAPTER TWO: LITERATURE REVIEW

2.0 Introduction

A great part of literature has been devoted to the study of the impact of economic forces on stock returns. In this chapter the theoretical and empirical issues will be discussed in order to have a better understanding of the effects of macroeconomic variables on stock returns.

2.1 Theoretical Review

In the finance literature, pricing of stocks has been a great concern. The Capital Asset Pricing Model (CAPM) and the Arbitrage Pricing Theory (APT) are the two models that have been extensively used to price financial assets and to measure the capability of those assets to generate a profit or loss.

2.1.1 Capital Asset Pricing Model (CAPM)

The Capital Asset Pricing Model describes the relationship between risk and expected returns, and serves as a tool to price risky securities. The model asserts that the only risk that needs to be taken into consideration is systematic risk because it cannot be eliminated through diversification. Investors should only be compensated for systematic risk assuming they hold a well-diversified portfolio. The non-diversifiable risk is represented by quantity beta (β) in financial theory.

The Capital Asset Pricing Model formula: Ri = Rf + βi (rm-rf)

Where:

Ri is the expected return by CAPM,

rf is the risk free rate,

rm is the market return and

βi is the risk factor

Main limitations of the Capital Asset Pricing Model.

It is based on unrealistic assumptions which do not apply to real world situations.

It is almost difficult to test the validity of CAPM.

Betas do not remain stable over time.

2.1.2 Arbitrage Pricing Theory

The Arbitrage Pricing Theory (APT) was founded by Stephen Ross in 1976. It is a multi-factor period model in which every investor believes that the stochastic properties of returns of capital assets are consistent with factors structure. Ross (1976) stressed that if equilibrium prices offer no arbitrage opportunities over static portfolio of assets, then the expected returns on the assets are approximately related to the factor loadings. In other words, the APT assumes that the return on asset is a linear function of several factors; company factors and macroeconomic factors or theoretical market indices, where the sensitivity to changes in each factor is represented by a factor-specific beta coefficient. This is highlighted in the formula below:

E(Ri) = Rf + β1iF1 + β2iF2 + β3iF3 + ⋅⋅⋅+βijFj

Where E(Ri) is the expected return of the underlying asset,

Rf is the risk free rate,

Each F is a separate factor and,

Each β measures the relationship between expected return and risk factors.

The APT initiated the use of macroeconomic variables by employing statistical tools like the factor analysis. The benefit of factor analytic techniques is that a large proportion of the risks in that particular dataset over the period are explained by the factors from the data. However the limitation of this method is that the factors usually have no economic interpretation.

The use of observed macroeconomic variables as risks factors is an alternative to factor analytic techniques. Chen et al. (1986) were the first to employ observed factors. They argue that at the most basic level some fundamental valuation model determines the prices of assets. In other words, the intrinsic value of a stock will be the correctly discounted expected future cash flows. Therefore the selection of factors should in include any systematic influences that impact on the future cash flows of corporates, the way investors form expectations and the rate at which they discount the cash flows.

2.1.3 The Arbitrage Pricing Theory Model

There is no need for investors to hold any particular portfolio. No special role is attributed to any market portfolio.

Only systematic or unsystematic risks matter, but there can be various macroeconomic factors that affect the returns of a well-diversified portfolio. It is up to the researcher to select the risk factors for example industrial production, interest rates, inflation, etc.

Investors must agree on what the relevant risk factors are. A linear relationship should be found between the risk exposure or sensitivity (its loadings on the risk factors) and the expected return of a security.

2.1.4 Assumptions of the Arbitrage Pricing Theory

There is competition in capital markets.

Investors always prefer more return to less with certainty.

The stochastic process which generates asset returns can be presented as a k-factor model.

2.1.5 Common Factor loadings in the Arbitrage Pricing Theory

The theory does not stipulate which factors should be employed for a particular stock or asset. In reality, one stock might be more sensitive to one factor than another. As a result the investor should identify each of the factors affecting a particular stock and the sensitivity of the stock to each of these factors. The following macroeconomic factors are said to play a significant role in influencing largely the return of a stock:

Inflation

GNP or Gross National Product

Confidence of investors as measured by surprises in default risk premiums for bonds

Industrial Production

Money Supply

Exchange Rates

2.2 Empirical Review

There are several empirical investigations in the literature which have applied Ross’s (1976) Arbitrage Pricing Theory model. Chen et al. (1986) performed the empirical study for the APT model in the US by selecting macroeconomic variables to estimate stock returns. Seven variables were used namely term structure, risk premium, industrial production, inflation, oil prices, market return and consumption during the period January 1953 to November 1984. Results revealed a strong relationship between the macroeconomic variables and the expected stock returns. They found that industrial production, changes in risk premium, twists in the yield curve, measure of unanticipated inflation of changes in expected inflation during periods when these variables are highly volatile, are significant explaining expected returns. Furthermore they also found that consumption and oil prices are not priced by the financial markets and conclude that asset prices respond sensitively to news pertaining to the economy, especially to unanticipated ones.

Altay (2003) employed factor analytic techniques to test the effect of macroeconomic variables on asset returns in the APT framework for both the German and Turkish economy. The dataset for the study was from January 1988 to June 2002 and January 1993 to 2002 for Turkey and Germany respectively. Eight macroeconomic variables were used for the study namely imports, exports, consumer price index, foreign exchange rate, average yield of public bonds, industrial production index, wholesale price index and money market interest rate. Results from the analysis showed that for the German stock market, he found evidence of only one factor beta which is the unexpected interest rate level factor beta. Quite surprisingly, the author did not found any macroeconomic factor beta with a significant influence on stock returns in the Turkish economy in the period considered.

Bundoo (2009) tested the Arbitrage Pricing Theory in Mauritius to analyze macroeconomic factors likely to influence the market return (SEMDEX return) on the Stock Exchange of Mauritius. The sample period was monthly observations from January 2002 to December 2006. Macroeconomic variables taken in consideration were oil price, tourist arrival rate, consumer price index, exchange rate, electricity consumption, aggregate money supply and Lombard rate. The study showed that four variables namely the level of the price index, the oil price (Mauritius depends heavily on oil imports), the exchange rate and electricity consumption (which is a proxy for the level of economic activity) were statistically significant at the 10% level or better in explaining variation in the equity premium in the SEM. Furthermore exchange rate was found to be the most important among the variables tested.

The development of cointegration analysis provided another approach to examine the relationships between macroeconomic variables and stock returns. A set of time-series variables are cointegrated if they are integrated of the same order and a linear combination of them is stationary. These kinds of linear combinations would then point to the existence of a long-term relationship between the variables. The strength of cointegration analysis is that through the construction an error-correction model (ECM), the dynamic co-movement among variables under consideration and the adjustment process toward long-term equilibrium can be examined, Maysami et al (2004).

The first group of studies covers developed countries. Mukherjee and Naka (1995) employed the Johansen cointegration tests in the Vector Error Correction Model (VECM) which covered 240 monthly observations for each variable from January 1971 to December 1990 to examine the dynamic relationship between inflation, exchange rate, industrial production, money supply, call money rate, long term government bond rate and the Japanese stock market. They document evidence that a long-term equilibrium relationship exists between the Japanese stock market and the six macroeconomic variables.

Using Johansen’s methodology for multivariate cointegration analysis and monthly time series data for the period January 1988 to January 1995, Maysami and Koh (2000) investigated the long-term equilibrium relationship between the Singapore stock index and some selected macroeconomic variables as well as stock indices of Singapore, Japan and the US. The study showed that industrial production and trade which are two measures of real economic activities were not integrated of the same order as changes in stock market levels. However changes in short- and long-term interest rate, money supply growth, inflation and variation in exchange rate do form a cointegrating relationship with changes in stock market levels in Singapore. Changes in interest and exchange rates have a significant impact in the cointegrating relationship, unlike price levels and money supply. This demonstrates that the Singapore stock market is interest and exchange rate sensitive. Furthermore with the building of a tri-variate model using the stock indices of Singapore, the U.S., and Japan, the study showed that the three markets are highly cointegrated.

Maysami et al. (2004) employed the Johansen and Jesilius (1990) protocols to investigate the long term equilibrium relationships between selected macroeconomic variables and the Singapore stock market index as well as the various Singapore Exchange Sector indices namely the property index, the finance index and the hotel index for the period of February 1995 to December 2001. The macroeconomic variables taken into account were interest rate, inflation, exchange rate, industrial production and money supply. Results showed that changes in the short and long term interest rates, price levels, industrial production, exchange rate and money supply influences significantly the Singapore’s stock market and the property index. This contradicts the findings of Maysami and Koh (2000), which found that price levels and money supply do not have a significant impact on the Singapore stock market.

Chaudhuri and Smiles (2004) applied the multivariate cointegration methodology in their study test the long run relationship between stock prices and changes in real macroeconomic activity in the Australian stock market in the period from 1960 to 1998. They found a long-run relationship between real stock price and the measures of aggregate real activity including real private consumption, real GDP real money and real oil price in the Australian market. The study also showed that stock returns variation in the US and New Zealand significantly affected movements in the Australian returns.

Employing Cointegration and Granger causality test, Gan et al. (2006) examined the relationships between the New Zealand’s stock market index (NZSE40) and a set of seven macroeconomic variables namely inflation rate, exchange rate, gross domestic product, money supply, long term interest rate, short term interest rate and domestic retail oil price for the period January 1990 and January 2003. Innovating accounting analyses have also been used to investigate the short run dynamic linkages between the New Zealand stock market index and the macroeconomic variables. Results from the Granger causality test shows that stock market index of New Zealand is not a leading indicator for changes in macroeconomic variables and the NZSE40 is consistently influenced by the rate of interest, money supply and real GDP.

Second group of studies investigate the relationship between macroeconomic variables and stock returns is for emerging countries. By employing standard and well-accepted methods of cointegration and vector auto-regression, Ibrahim and Aziz (2003) analyzed the dynamic relationships between stock prices and four macroeconomic variables namely industrial production, exchange rate, money supply, price level and the Malaysian equity price. The data are monthly observations for the period from January 1977 to August 1998. Empirical results showed the existence of a long-run relationship between these variables and stock prices. Substantial short term interactions have also been found. Exchange rate is negatively related to stock prices. For money supply there is immediate positive liquidity effects and negative long-run effects of money supply expansion on the stock prices.

Acikalin et al. (2008) employed cointegration tests, a vector error correction model (VECM) and causality tests to examine the relationship between GDP, exchange rates, interest rates and current account balance and stock market returns in Turkey from the last quarter of 1991 to the last quarter of 2006 . The cointegration tests and VECM showed a direct long-term equilibrium relationship between the set of macroeconomic variables and stock price index. Current account balance, exchange rates and GDP have negative impacts on stock market returns. The Granger causality tests showed uni-directional relationships between macroeconomic variables and ISE index. However, contrary to the existing literature, stock market does affect interest rates but not the other way around.

Anokye and George (2008) analyzed the relationship between macroeconomic variables and the stock market in Ghana with quarterly data ranging from 1991 to 2006. Inflation, interest rate, exchange rate, oil prices and inward FDI were taken into consideration. Using Johansen's multivariate cointegration test and innovation accounting techniques, results showed that there was significant relationship between the variables. The presence of a cointegration relationship between the variables and stock prices is a signal that stock market efficiency is in doubt. Interest rate and foreign direct investment are the key determinants of stock price movements as demonstrated by the Impulse Response Function (IRF) and Forecast Error Variance Decomposition (FEVD).

Jiranyakul (2009) explored the relationship between the Thai stock market and a combination four macroeconomic variables namely real GDP, money supply, nominal effective exchange rate and the stock index from the first quarter of 1993 to the fourth quarter of 2007. The Johansen cointegration test showed cointegration among the variables whereas the Engle and Granger test does not exhibit cointegration. Two cointegrating vectors were found. Real GDP, money supply and nominal effective exchange rate have a significant impact on the stock market index. On the other hand price level imposes a negative impact. Bidirectional causality exists between stock market and economic growth. It should be noted that the crisis imposes no impact on long run relationship.

Kosnandi (2011) investigated the relationship between macroeconomic factors and the FTSE Bursa Malaysia stock markets price index for the period of 2000 to 2009 as monthly. The macroeconomic variables consisted of inflation rate, interest rate, money supply and exchange rate. By employing Johansen’s multivariate cointegration, the existence of long term relationships between all variables has been found. Furthermore each variable is cointegrated with each other. Results from the Granger Causality test revealed that for inflation, interest rate and exchange rate, there is one-way causality with stock market index. On the other hand bi-directional causality can be observed between money supply and the stock market. The main limitation of the study is the use of only four macroeconomic variables.

Using multivariate cointegration analysis, Heista (2011) tested the short run and the long run relationship between selected macroeconomic variables and the stock exchange of Mauritius. For the study quarterly data from 1993 to 2010 has been used. Variables taken into account were economic activity (proxied by the level of trade), inflation (proxied by consumer price index), oil prices, exchange rates, aggregate money supply and overall average weighted yield on treasury bills. The Johansen’s results revealed three cointegrating relationships among the variables. Results from the granger causality test shows that inflation, money supply, oil prices and exchange rates influence stock returns. On the other hand, stock returns granger cause yield on treasury bills.

There is another group of studies which examines the situation for more than one country. Cheung and Ng (1998) concluded that there are long term co-movements between the national stock index and some specific variables like real consumption, real money supply, real oil price and real GNP output in Canada, Germany, Italy, Japan and US. They applied the Johansen’s technique with quarterly data. In addition the authors found that real returns on stock indexes are related to deviations from empirical long term relationships and to changes in macroeconomic variables.

Using the Johansen Cointegration tests in their studies, Nasseh and Strauss (2000) found a significant, long-run relationship between stock returns and both domestic and international economic activities in six European countries. The domestic variables consist of industrial production, surveys of manufacturing orders, short and long-term interest rates, while the international variables were made of foreign stock returns, short term interest rates and production. Variance decompositions were also used to ‘support the strong explanatory power of macroeconomic variables in contributing to the forecast variance of stock returns".

Wongbang and Sharma (2002) explored the relationship between the stock returns and macroeconomic variables for the ASEAN-5 countries consisting of Malaysia, Indonesia, the Philippines, Singapore and Thailand. The variables taken in account were gross national product (GNP), consumer price index (CPI), money supply, interest rate and exchange rate. By observing both short and long run relationships between the respective stock indexes and the selected variables, they found that in the long-run all five stock price indexes were positively related to growth in output and negatively to the price level. On the other hand a negative long-run relationship was noted between stock prices and interest rates for Philippines, Singapore and Thailand and was found to be positive for Indonesia and Malaysia. The causality tests showed an overall relationship between stock prices and macroeconomic variables for all five ASEAN markets.

Humpe and Macmillan (2007) examined the long term movements caused by macroeconomic variables. They made a comparison between US and Japan. The macroeconomic variables that were taken into account were consumer price index, industrial production, money supply, rate of interest and monthly data over the last 40 years were employed. Findings showed that in US stock prices react positively to industrial production. Inflation and rate of interest negatively affect stock prices while money supply had practically no effect. In Japan, two cointegrating vectors were depicted. Stock prices responded positively to industrial production and negatively to money supply. The second cointegrating vector showed that industrial production is negatively related to consumer price index and interest rate.

Mahmood and Dinniah (2009) examined the dynamic relationship between stock prices and economic variables in Malaysia, Korea, Thailand, Hong Kong, Japan and Australia. The study used monthly data from January 1993 to December 2002 and the selected macroeconomic variables were consumer price index foreign exchange rates and industrial production index. Using a multivariate approach, the results found a long run equilibrium relationship between stock price indices and among variables in only four countries, that is, Japan, Korea, Hong Kong and Australia. Concerning the short run relationship, some interactions were found for all countries except for Hong Kong and Thailand. Hong Kong shows relationship only between exchange rate and stock prices and there is significant interaction only between output and stock prices in Thailand.

Hosseini et al. (2011) investigated the relationships between four macroeconomic variables namely money supply, crude oil price, inflation rate and industrial production in China and India covering the period January 1999 to January 2009. The Johansen-Juselius Multivariate Cointegration and Vector Error Correction Model technique showed that there are both long and short run linkages between the selected macroeconomic variables and stock market index in both economies. In the long run, increases in crude oil prices in China are positive but the effect is negative in India. For the case of money supply, the impact for the Indian stock market is negative whereas for China, there is positive impact. Industrial production has a negative influence only in China. Finally inflation has a positive influence on stock market indices in both countries.

CHAPTER 3: DEVELOPMENT OF THE STOCK EXCHANGE OF MAURITIUS

3.0 Introduction

After the first phase of modernization in the late 1980s, Mauritius experienced the need for a well-structured financial sector. The aims were to develop new avenues for progress, intensify the level of competition in the financial system for an efficient resource allocation in the economy. Institutional and policy reforms were the key to phase the financial infrastructure and liberalise the trade in financial services. With the recommendations of the World Bank and International Monetary Fund (IMF), Mauritius reviewed its regulatory framework, banking supervision and aimed at fiscal discipline accompanied with macroeconomic stability. Moreover, there was a need for the enhancement of the capital market to enable investors and savers to trade financial instruments. The incorporation of the SEM in 1989 has tremendously helped Mauritius to be ranked among the top countries in the Africa.

The Stock Exchange of Mauritius was set up in Mauritius on March 30, 1989 under the Stock Exchange Act 1988, as a private limited company with the main objective to operate and promote an efficient and regulated securities market in the country. The SEM has become a public limited company on October 6th, 2008 and in order to keep pace with the ever transforming world of international stock markets and to remain a favourable destination for investors, the exchange continuously review and update its operational, regulatory and technical framework when required so as to mirror the ever-changing standards of the stock market environment around the world. Since its inception in 1989, the SEM has grown to be one of the leading stock markets in Africa in terms of stock market infrastructure and operational efficiency. SEM is ranked as the 5th exchange by market capitalisation in sub-Saharan Africa and as being one of the leading frontier African markets.

In November 2005, the SEM has become a full member of the World Federation of Exchanges. The attainment of membership status of the World Federation of Exchanges represents an important milestone that have allowed the SEM to be in the league of stock markets that are compliant with the stringent standards and market principles established by the WFE. It should be noted that the WFE is a reference point and set standards for stock markets worldwide. Furthermore the SEM is accredited to international bodies like the South Asian Federation of Exchanges (SAFE), African Securities Exchanges Association (ASEA) and Committee of SADC Stock Exchanges (COSSE) and is recognized by the UK Her Majesty’s Revenue & Customs (HMRC). For the past few years, the SEM has gone through a strategic reorientation of its activities and is trying to change from a predominantly equity-based domestic stock exchange to a multi-product internationally oriented stock market so as to provide Mauritius a financial architecture which will keep pace with the modernization of the world economy.

The SEM operates two markets: the Official Market and the Development & Enterprise Market (DEM). In 1989, with five listed companies and a market capitalization of around USD 92 million the Official Market started it operations. Currently, 41 companies are listed on the Official Market which represents a market capitalization of approximately USD 5.7 billion as at 31 December 2012. On the other hand, the DEM was established on the 4th of August in 2006 and is aimed at small and medium sized enterprises and newly set-up companies who want to benefit from the facilities given by an organized and regulated market in order to raise capital for future growth. As at 31 December 2012, there are 47 companies listed on this market. The market capitalization of the DEM is nearly USD 1.4 billion. The companies listed on the Official Market and the DEM are classified into eight categories namely Banking, Insurance and Other finance, Commerce, Industry, Investments, Leisure and Hotels, Sugar, Transport, Foreign.

Figure 1: Total value traded in the Official market

Source: Stock Exchange of Mauritius Annual Report 2012.

Figure 2: Total value traded in the Development & Enterprise Market

Source: The Stock Exchange of Mauritius Annual Report 2012

Table 2: Major changes that have marked the SEM in the last two decades

1989

2009

Number of companies listed

(Official + DEM)

6

89

Market capitalisation

(Official + DEM) (Rs bn)

1.44

175.1

Market capitalisation as a % of GDP

2.5

66.4

Semdex

117.34

1660.87

Trading System

Manual

Automated Trading (ATS)

Internet Trading

Nil

I-Net

Disclosure Requirements

Limited

Extensive

Reporting Cycle

Half-Yearly

Quarterly

Number of trading sessions

26

250

Source: Annual Report 2009 of the Stock Exchange of Mauritius

3.1 The Main Indicators of Growth for the Stock Exchange of Mauritius

Table 3: Selected Market indicators for the Stock Exchange of Mauritius

Year

2004

2005

2006

2007

2008

2009

2010

2011

Market Size indicators

Market capitalisation

(Rs billion)

67.03

80.04

116.98

173.09

109.30

151.21

178.0

171.51

Listed Companies

(End of Period)

40

41

41

41

40

40

37

38

Change in Market Capitalisation (%)

30.9

19.4

46.16

47.97

-36.86

38.35

17.71

-3.64

Market Cap/GDP (%)

38.2

43.18

56.96

74.93

41.42

54.59

61.40

52.81

Liquidity Indicators

Turnover/GDP (%)

1.61

2.45

2.92

5.12

4.32

3.78

4.06

4.61

Turnover/Market Cap (%)

4.21

5.68

5.12

6.83

10.43

6.93

6.61

8.72

Source: The Stock Exchange of Mauritius Factbook 2012

The market capitalisation shows the overall size of a stock market. As can be seen in table above there has been a gradual increase in market capitalisation from 2004 to 2007. This indicates the progress that the SEM has made during those years. However in 2008 a large negative drop of 36.86% had occurred which enlightens the fact that the US Subprime crisis has adversely affected SEM during that particular period. Furthermore we can notice that the government has made significant effort to contain the effects of the financial crisis through its stimulus package as this is reflected by a 38.35% rise in 2009.

Liquidity is crucial because it shows that if stock markets are liquid, this helps to improve the allocation of capital and enhance the prospects of economic growth. The turnover ratio is the value of shares traded divided by the average market capitalisation. This ratio complements the ratio of turnover to GDP, because the turnover ratio encompasses the size of the market and the value traded ratio of the economy. Significant increase can be seen in the turnover ratio from 2004 to 2007 which shows that the level of trading activities were low. With the introduction of the Central Depository & Settlement Co. Ltd (CDS) in 1997, this has brought quick and efficient clearing and settlement of trade. Eventually the CDS has raised the level of liquidity to another level which enables market participants to buy and sell securities in just a matter of time. In 2008, there has been a hike in the turnover ratio because of capital flight. The Subprime Mortgage Crisis induced investors to sell their liquid stocks as they have lost confidence in the market. Thereafter the turnover ratio has gained back to its pre-crisis level.

3.2 The Performance of the SEM vis-à-vis Emerging Stock Exchanges in Africa

The SEM has made remarkable progress during the last two decades and is considered as a solid international platform in the African continent. The main question remains that to which extent the SEM has been progressing compared to other emerging stock markets in Africa. With the liberalization of emerging economies and their removal of foreign capital controls, capital flows has risen continuously to those market. Kehl (2007) argued that despite harbouring substantial risks, emerging markets give investors opportunities for reaping high returns.

The Stock Exchange of Mauritius was awarded for the "Most Innovative African Stock Exchange of the Year Award" at the Africa investor (Ai) prestigious annual Index Series Awards in September 2012. The criteria for this Award category were initiatives put in place by the exchange to exploit new areas of development, programmes implemented to improve the services it provides to its key shareholders and compliance of the exchange’s regulatory and operational set-up with international standards. In 2007 the SEM was the runner up of the "Best African Stock Exchange" category. However there is a long way to go to reach become a success story like the Johannesburg Stock Exchange in terms of liquidity, market size and market capitalisation which is considered as the leading stock market in Africa.

Botswana Stock Exchange, Ghana Stock Exchange and Nairobi Stock Exchange were chosen for a comparative status with the Stock Exchange of Mauritius. They were selected on based on data availability, market size and market capitalisation. Botswana, Ghana and Kenya are considered as emerging markets and have attracted the attention of international investors. Botswana is the third largest stock exchange in terms of market capitalization in Southern Africa and is one of the best performing stock exchanges with annual returns averaging 24% in the past decade. Ghana has made significant progress during the last ten years and has achieved macroeconomic stability in term of GDP growth. This can be attributed to the emergence of majors sectors of the economy including the money markets (financial institutions) and the capital markets (debt and equity). In Kenya, the Nairobi Stock Exchange has grown since in inception as an avenue for the mobilization of resources. International and local investors have benefited of the facility as a provider of investment opportunities. The Nairobi Stock Exchange has been a pre-eminent bourse in the eastern and central region of Africa. Table 4 below gives a comparison between SEM and those three markets.

Table 5: Comparison of the SEM with other Emerging African Stock Exchanges

Stock Exchange/Year

Number of firms listed

Mkt Cap

(US$ billion)

Mkt Cap (% of GDP)

Turnover (% of GDP)

Turnover ratio

Stock Exchange of Mauritius (Mauritius) - Founded in 1989

Market Size

Indicators

Liquidity

Indicators

2009

89

4.74

53.65

3.73

8.06

2010

86

6.51

67.03

3.68

6.36

2011

86

6.54

58.08

4.64

8.01

Botswana Stock Exchange (Botswana) - Founded in 1989

2009

20

4.27

37.08

0.89

2.64

2010

21

4.08

27.35

0.94

3.35

2011

23

4.11

23.70

0.84

3.55

Ghana Stock Exchange (Ghana)- Founded in 1989

2009

35

2.51

9.65

0.22

1.96

2010

35

3.53

10.98

0.32

3.37

2011

36

3.09

7.90

0.35

4.13

Nairobi Stock Exchange (Kenya)- Founded in 1954

2009

55

10.8

35.17

1.62

4.59

2010

55

14.5

44.91

3.38

8.66

2011

58

10.2

30.34

2.61

7.12

Source: World Development Indicators 2012

As can be observed, Mauritius does pretty well with respect to Botswana, Ghana and Kenya. The market capitalisation as a percentage of GDP measured in US dollars is higher as compared those three markets. We can point out that the SEM has progressed significantly with respect to Botswana and Ghana since the establishment of the three exchanges in 1989. Furthermore the number of listed firms in the SEM is superior to that Botswana, Ghana and Kenya. The liquidity indicators show that Mauritius is well stands above its counterparts. The figures indicate that the SEM performs better vis-à-vis its partners. Thus the trading activity was quite high in the SEM as compared to these markets. One major observation is that despite having a market size superior than the SEM, the Nairobi Stock Exchange is not liquid enough and is characterized by low trading activities and is being inactive.

3.3 Market Indices

SEMDEX

The SEMDEX is an index of share prices of all listed companies in the official market and each stock is weighted with respect to its share in the total market capitalisation. Any movement in the semdex indicates changes in the share prices of companies with higher market capitalisation. In the calculation of semdex, the base period is 5th July 1989 with an index value of 100.

The formula is as follows:

SEMDEX = Current Market Value of All Listed Shares x 100

Base Market Value of ALL Listed Shares

It should be noted that the market value of any class of shares is equal to the number of outstanding shares multiplied by its market price. In order to reflect new listings, rights issues and other capital restructurings, the base value of listed shares is adjusted accordingly.

SEMTRI

The SEMTRI is a Total Return Index with the main aim to provide domestic and foreign market participants a vital tool to analyse the performance of the local market. Like the already published all-share index SEMDEX, the SEMTRI captures the price movements of listed stocks. Additional features are incorporated in the Total Return Index to enable investors, particularly long term investors such as pension funds to have an insight of the total return which combines gains/losses on the listed stocks and gross dividends obtained on these stocks since the setting up of the local market in 1989.

SEM-7

The SEM-7 was established in 1998 and it is made up of seven companies listed in the official market which have the largest market capitalisation. Liquidity and investibility of shares are the most important criteria in order to be included in the index. There is an Index Management Committee which ensures the proper maintenance of the index.

Table 6: The Evolution of Market Indices since 2004 (In Rupee Terms)

Year

2005

2006

2007

2008

2009

2010

2011

SEMDEX

804.03

1204.46

1852.21

1182.74

1660.87

1967.45

1888.38

SEM-7

175.43

264.41

477.40

267.22

360.75

373.22

350.33

SEMTRI

1951.83

3060.71

4868.61

3233.74

4712.70

5747.85

5673.88

Source: SEM Factbook 2012

Table 7: Selected International Market Indices – Comparative Indicators

Indices

2002

2011

Compound Annual Growth Rate (%)

SEMDEX (Mauritius)

340.92

1888.38

18.7

Bovespa (Brazil)

13577.00

567754.00

15.4

S&P/ CNX 500 (National Stock Exchange of India)

1059.05

3597.75

13.0

FTSE/JSE All Share (South Africa)

10441.70

31985.67

11.8

All Ordinary Price (Australian SE)

3359.90

4111.00

2.0

Dow Jones (NYSE)

10021.57

12217.56

2.0

CAC-40 (France)

4624.58

3159.81

-3.7

Source: The Stock Exchange of Mauritius

Conclusion

In this chapter, we have demonstrated the progress of the Stock Exchange of Mauritius vis a vis the emerging countries in Sub Saharan Africa. Furthermore some statistics were provided to capture the drastic evolution of the SEM which occurred over the recent years. We can san say that the SEM is an important institution in the Mauritian economy through its contribution to GDP and inflow of foreign direct investment and portfolio investment.

RESEARCH METHODOLOGY

Model Specification

The estimation model is constructed before the econometric analyses of the variables selected are carried out. Money supply, exchange rate, oil prices, overall yield on treasury bills and consumer price index are used to assess their influence on the stock market returns. All data collected for the study will be measured in the form of logarithm in order to obtain a more accurate result and to avoid different units of measurement for each variable.

Expressing the equation in logarithmic form:

LSEDX = β0 + β1LM2 + β2LEXC + β3LOIL + β4LYIELD + β5LCPI + εt

Where,

LSEDX = Stock market price index

LM2 = Money supply

LEXC = Exchange rate

LOIL = Oil prices

LYIELD = Overall yield on treasury bills

LCPI = Consumer price index

β = intercept

βi = estimated coefficients; i = 1, 2, 3…

εt = error term

Description of data

Semdex

It is an index of prices of all listed companies. Each stock is weighted in proportion to its share in the total market capitalisation. Changes in Semdex are caused by movement in the prices of shares with relatively high market capitalisation.

Aggregate money resources (Broad Money)

In Mauritius, the central bank defines money supply as comprising of narrow money (M1) and broad money (M2). Narrow money comprises of currency and coins whereas broad money is made up of savings, small time deposits, overnight repos at commercial bank and non-institutional money market accounts. Broad money measures the total volume of money supply in the economy. The excess liquidity may arise in the economy when the amount of broad money is over and above the level of output in the economy.

Consumer Price Index (CPI)

The CPI is an indicator of changes over time in the general level of prices of goods and services acquired by Mauritian consumers. It measures the price change by comparing, through time, the cost of a fixed basket of goods and services.

Exchange Rate

The nominal exchange rate used is the indicative Rs/US exchange rate because banks in Mauritius first adjust the rupee against the US dollar, taking into account daily movements in the US dollar on the international market and the liquidity situation of the foreign exchange market, before crossing the rate thus obtained against other currencies according to their movements on the international market.

Oil prices

Oil is an important component in the production of goods in Mauritius. Oil powers industries, heats buildings, and provides the raw material for plastics, paints, textiles, and in transportation. Given Mauritius is a net oil importer, fluctuations in the prices of oil have an impact in the cost of the firms and hence their discounted cash flows.

Treasury Bills

Treasury bills are a source of funds for the government in Mauritius to make budget shortcomings or required investment. We take the overall average weighted yield on 28, 91, 182, 364 and 728 Day treasury bills.

The study uses a total of five macroeconomic variables and the stock market index for the analyses which are described in the table below:

Table 2:

Variables

Concept

Definition

Source

LSEDX

Log of stock returns

Semdex: stock returns

SEM Factbook

LM2

Log of money supply

Broad money supply M2

Bank of Mauritius

LEXC

Log of exchange rate

Exchange rate of rupee vis a vis US dollar

Bank of Mauritius

LOIL

Log of oil prices

End of month retail oil prices in terms of US dollar.

World Bank

LYIELD

Log of treasury bills

Treasury bill rate overall weighted yield-91,182,364 days

Bank of Mauritius

LCPI

Log of consumer price index

Consumer price index– measure of inflation

Bank of Mauritius

(Source: computed)

The study employed secondary data obtained from monthly bulletins of the Bank of Mauritius, the SEM Factbook and the World Bank. For the stock returns, end of month values for SEMDEX are taken as proxy. The macroeconomic variables are monthly frequencies from June 1998 to June 2010 leading to a total of 145 observations for each variable. The natural logarithms of the variables are taken to measure the elasticity of SEMDEX with respect to each macroeconomic variable. In other words the percentage change in SEMDEX for a given percent change in each macroeconomic variable.

Data Analysis

Time series data will be analyzed using the econometric software STATA version 10.0. The unit root test will be applied to examine the stationarity of variables. The Johansen and Juselius (1990) cointegration test will be employed to test the long run relationship between the dependent variable and the independent variables and to find the number of cointegrating vectors. Furthermore the Granger causality test will be run to determine the causality direction between variables.

Test for Stationarity- Unit Root Test

The unit root test is conducted before an analysis of cointegration. Unit root test is based on testing the null hypothesis against the alternative hypothesis of stationarity. Several studies have demonstrated that macroeconomic time series are non-stationary. In line with the conditions of non-stationarity data, the normal properties of Durbin-Watson (DW), t statistics and the measure of R2 will no longer hold and this will lead to spurious results in case we perform the regression. In order to test for stationarity, the Augmented Dicker Fuller (ADF) test will be employed.

Augmented Dicker Fuller Test

The ADF test is used to investigate the relationship between dependent and independent variables in the long run. An ADF model can be shown as follows:

Where εt is a white noise error term, Yt = {LSEDX, LM2, LEXC, LYIELD, LOIL, LCPI} which is the series being tested, ∆ is the difference operator and t is the time trend.

The hypothesis under the ADF test:

H0: δ = 0

H1: δ < 0

The null hypothesis is that δ = 0. The time series is stationary when the null hypothesis is rejected. The null hypothesis of δ = 0 will be tested against the alternative hypothesis δ < 0. If the series is not stationary at level form, it will be differenced d times to be stationary to determine the order of integration.

Cointegration Test

Cointegration is employed to know whether the model has cointegrating vectors or not. In other words cointegration examines the long-run relationship between macroeconomic variables and stock returns. In most cases if two variables are integrated of order one, that is I(1), their linear combination will also be stationary. Granger (1986) stated that a cointegration test acts as a pre-test to avoid spurious regression circumstances. For the cointegration analysis, the Maximum Likelihood estimation method of the Johansen and Juselius (1990) multivariate approach will be used. This test stipulates that the rank of the matrix is equal to the number of cointegrating vectors and it also determine the long run relationship between the explained and explanatory variables.

Johansen and Juselius Procedure

The Johansen method is a procedure for testing cointegration of several variables which are integrated of order one. This test allows more than one cointegrating relationship.

Assume that a set of g variables are I(1) and are said to be cointegrated. It is possible to construct a VAR including k lags having these variables.

Yt = β1 yt-1 + β2 yt-2 + … + βk yt-k + ut

Where,

Yt is g x 1, matrix of non-stationarity.

βi is g x g matrix of parameters.

In order to apply the Johansen test, the VAR should be reformulated into a Vector Error Correction Model (VECM) as shown below:

∆yt = Г1∆yt-1 + Г2∆yt-2 +…+Гk-1∆yt-k+1 + ∏ yt-k + ut

Where,

Гi = -(I - β1-…-βi), I = 1,…,k-1 and ∏ = -(I - β1 -…-βk).

Information about the short and long-run adjustment to changes in yt via estimates of Г and ∏ respectively is specified in the system. This VAR contains g variables in the first differenced form on the left hand side and k-1 lags of the dependent variables (differences) on the RHS, each with a Г connected to it. The Johansen test is very sensitive to the chosen length lag; therefore it is important first to determine the appropriate lag length before using testing for cointegration.

The Johansen test focuses on the examination of the ∏ matrix. It can be defined as the long run matrix because in equilibrium, all ∆yt-i will be zero and by making the error terms ut , equal to their respective expected value of zero, we will get ∏ yt-k = 0.

∏ = αβ’, where α denotes the speed of adjustment to disequilibrium and β is a matrix of long run coefficients.

Since ∆yt…∆yt-k+1 are all I(0) but yt is I(1) while ∏ yt-k must be stationary for ut to be I(0), there are three possible cases where the condition ∏ yt-k ~ I(0) can be satisfied:

When are variables in the system are I(0) which means there is full cointegration, that is Rank ( ∏ ) = p. A simple VAR is the suitable model in this case.

When Rank ( ∏ ) = 0 implying no cointegration among the variables in the long run. The appropriate model will be a first difference VAR.

The last way for the condition to be satisfied is when ∏ has a reduced rank of r ≤ (n-1). In other words, there exists up to (n-1) cointegrating relationships: β’ yt-k ~ I(0). In this case the existence of r ≤ (n-1) cointegrating vectors can be noted in β with (n-r) non-stationary vectors. Only the cointegrating vector in β enter the VECM implying that the last (n-r) are effectively zero

Johansen and Juselius (1990) specify two test statistics for cointegration which can be shown in equation (i) and (ii):

Trace Test: Ttrace =

Where, T is the total number of observation, N is the number of variables and ri is the i-th pair of variables. T trace has a chi-square distribution with N-r degrees of freedom. Large value of T trace gives evidence against the hypothesis of r or fewer cointegration vectors.

Maximum Eigenvalue Test: Tmax = -T ln (1 – λr+1)

The maximum eigenvalue test evaluates the null hypothesis of H0: r = r0 against H1: r = r0+1. The null hypothesis of r cointegration vectors id tested against the alternative of r + 1 cointegrating vectors.

Selection of the lag length

Before proceeding to the Johansen procedure, the determination of the lag length of the VECM is important. The lag length should be small enough in order to allow estimation and high enough to guarantee that the errors are approximately white noise. An insufficient lag length can result in the rejection of the null hypothesis of no cointegration and on the other hand over-parameterization of the lag length may lead to loss of degrees of freedom. The Johansen test employs the information multivariate criteria namely the Akaike information criterion (AIC) and Schwarz Bayesian information criterion (SBC).

The Akaike information criterion is a gauge of the goodness of fit of a model. It emphasizes the tradeoff between bias and variance in the model construction, or that of accuracy and complexity of the model. It also allows for comparison in the model.

The AIC is defined as

AIC= 2K-2ln (L)

Where k is the number of parameters in the statistical model and L is the maximized value of the likelihood function for the estimated model.

The Schwarz Bayesian criterion is a criterion for the model selection among a set of parametric models with different numbers of parameters. The problem of over fitting arises while estimating; hence SBC solves this constraint through a penalty (larger than the AIC) term for the number of parameters in the model.

The formula for SBC is as follows:

SBIC= -2Ln L+ KLn (n)

Where k = no of parameters, n is the number of data points in the observed data and L is the maximized value of the likelihood function for the model under estimation.

Vector Error Correction Model

An error correction model is a dynamic system with the characteristics that the deviation of the current state from its long-run relationship will be fed into its short-run dynamics. The VECM is a full information maximum likelihood estimation model and it yields more efficient estimators of cointegrating vectors. Without requiring a specific variable to be normalized, the VECM enables the testing for cointegration in a whole system in one step.

The general form of a VECM is shown below:

∆Yt =

Where, and are the components of the vector autoregression in first differences and error correction components. Yt represents a p x 1 vector of variables and is integrated of order one. μ is a p x 1 vector of constants, k is the lag structure and ut is a p x 1 vector of white noise error terms. Short run parameters are represented by Гj which is a p x p matrix of these coefficients across p equations at the jth lags. β is a p x r matrix of cointegrating vectors and α represents the adjustment parameters which is the speed of error correction mechanism. ∆ is the difference operator.

Granger Causality Test

Granger causality test is employed to analyze the causality direction between stock market index and its selected determinants. The causality test requires that all data series into observation should be stationary. This test is carried out to see the short-run causality running from independent variables to the dependent variable.

The Chi Wald test is used to examine the causality between the dependent variable and the explanatory determinants. The hypothesis is shown below:

H0: One variable granger does not granger cause the other

H1: One variable granger cause the other

The decision rule is that if the chi statistic is less than the chi critical value, we do not reject H0 and if the chi statistic is greater than the chi critical value we reject H0.

CHAPTER 5: ANALYSIS AND FINDINGS

Table 1: Descriptive Statistics: June 1998 to June 2010

Variables

Observations

Mean

Std deviation

Minimum

Maximum

LSEDX

145

6.558

0.572

5.831

7.603

LM2

145

25.719

0.586

24.972

30.425

LEXC

145

3.373

0.0881

3.195

3.544

LYIELD

145

2.135

0.335

1.247

2.575

LOIL

145

3.659

0.596

2.343

4.887

LCPI

145

4.769

0.0748

4.636

4.933

∆ LSEDX

145

0.00877

0.053

-0.195

0.226

∆ LM2

145

0.01006

0.544

-4.587

4.609

∆ LEXC

145

0.00208

0.022

-0.081

0.117

∆ LYIELD

145

-0.00666

0.065

-0.197

0.232

∆ LOIL

145

0.01242

0.091

-0.311

0.201

∆ CPI

145

0.00062

0.034

-0.291

0.029

(Source: Computed)

The above table depicts the descriptive statistics of the variables. As can be seen the standard deviation of Semdex is 0.572 which indicates that the Stock Exchange of Mauritius is quite a volatile market. Furthermore the average monthly return of semdex shows a fair average of 0.87% which is equal to an annualized return of 10.44% per year with a standard deviation of 5.3%. The Semdex earns maximum return of 22.6% in one month and maximum loss of 19.5%.

The Unit Root Test

The first step of the analysis is to check the presence or absence of stationarity in the time-series data by using the Augmented Dicker Fuller (ADF) test. The appropriate lag length for each variable is determined prior to performing the test. As a matter of fact, for cointegration analysis to be valid, the time series data should be integrated of the same order. In most cases the series must be integrated of order one. The tables below summarize the results of the ADF test.

Table 2: Unit root test results at level form

Series

ADF Test Statistic

Lag-Length

p-value

Conclusion

LSEDX

-0.648

4

0.8547

Non-Stationary

LM2

-1.839

4

0.3615

Non-Stationary

LEXC

-1.958

1

0.3055

Non-Stationary

LYIELD

-0.772

2

0.8273

Non-Stationary

LOIL

-1.851

2

0.3557

Non-Stationary

LCPI

-2.950

1

0.0398

Non-Stationary

(Source: computed)

Table 3: Unit root test results at first difference

Series

ADF Test Statistic

Lag-Length

p-value

Conclusion

LSEDX

-4.118

3

0.0009

I(1)

LM2

-8.848

4

0.0000

I(1)

LEXC

-7.760

1

0.0000

I(1)

LYIELD

-5.663

2

0.0000

I(1)

LOIL

-5.713

2

0.0000

I(1)

LCPI

-8.500

1

0.0000

I(1)

(Source: computed)

The results show that at level form all variables are non-stationary since their respective test-statistic does not exceed its critical value. Therefore the null hypothesis of a unit root cannot be rejected and any standard regression will produce spurious results. However, when the first difference was done, all variables were found to be stationary. Stock market index, money supply, exchange rate, overall weighted yield on treasury bills, oil prices are found to be stationary at 5% significant level and consumer price index at 1% significant level. This indicates that all variables are integrated of order one. We can say that there is a possibility of a long run relationship among the variables. In order to find such a relationship we proceed to the Johansen Multivariate test.

Cointegration Test Results

Before proceeding to the cointegration test, it is important to determine the optimal lag length. The model lag selection was determined by the Schwarz Information Criterion (SIC) and the Akaike (AIC) Information Criterion. An optimal lag length is vital because it avoids the problem of serial correlation. The objective will be to select the number of parameters which minimizes the information criteria. The drawback of SIC is that it tends to underestimate the lag order, while adding more lags will increase the loss of degree of freedom. The AIC is chosen as a leading indicator in order to make sure that there is no remaining autocorrelation in the VAR model. Since monthly frequencies are used in the study, Ivanov and Kilian (2001) mentioned that AIC tends to be more accurate with monthly data. Therefore the optimal lag length selected is 2. The table below shows the results for the selection of the appropriate lag length

Table 4: VAR lag order selection by information criteria

Lag

AIC

SBIC

0

-2.04432

-1.91884

1

-15.3857

-14.5073*

2

-15.4818*

-13.8506

3

-15.37

-12.9859

4

-15.041

-11.904

(Source: computed)

The Johansen and Juselius (1990) procedure is useful in determining whether there exists a long-run relationship among the variables. Table 5 estimates the number of cointegrating vectors which exists between stock market index and the macroeconomic variables.

Table 5: Results from Johansen’s Cointegration Test (Trace and Maximum Eigenvalue)

Trace Test

Null hypothesis

Alternative Hypothesis

Eigenvalue

Trace Statistic

5% critical value

r = 0

r = 1

126.7153

94.15

r ≤ 1

r = 2

0.38030

58.2865*

68.52

r ≤ 2

r = 3

0.17259

31.1943

47.21

r ≤ 3

r = 4

0.10633

15.1178

29.68

r ≤ 4

r = 5

0.06635

5.3007

15.41

Max-Eigen Test

Null Hypothesis

Alternative hypothesis

Eigenvalue

Max-Eigen Statistic

5% critical value

r = 0

r = 1

68.4287

39.37

r ≤ 1

r = 2

0.38030

27.0923

33.46

r ≤ 2

r = 3

0.17259

16.0765

27.07

r ≤ 3

r = 4

0.10633

9.8170

20.97

r ≤ 4

r = 5

0.06635

5.2052

14.07

(Source: computed)

As can be observed in the table above, both the Trace statistic and Max-Eigen statistic indicate the presence of one cointegrating vector. In other words the series has a cointegrating rank of one. We reject the null hypothesis of no cointegrating equilibrium at the 5% significance level and conclude that there is a long run relationship between stock market returns and macroeconomic fundamentals. The Johansen Test suggests two main assertions. First, in the long run the variables move together and short term deviations will be corrected towards equilibrium. Secondly, when there is cointegration this indicates causality in at least one direction

Since one cointegrating vector has been found, the relationship between stock market index and the macroeconomic variables cannot be modeled in a VAR. In this case, a VECM is the most appropr