The History Of Capital Asset Pricing Model

Published Date: 02 Nov 2017

HAPTER 1

INTRODUCTION

1.0 INTRODUCTION

An efficient Stock Market projects the strength and the health of a country’s economy. A truly efficient stock market can assist the development process of a country through two channels by boosting savings and allocating for a more efficient distribution of resources. The basic sense of stock market serves as a platform for the transfer of funds from the deficit to the surplus side. Furthermore, it not only serves in the transfer of wealth but also in the sharing of risk. Stock market is an important institution in a country and is of great concern to investors, stakeholders and the policy markets; it is essential in the sense that it helps capital formation and economic growth.

1.2 PROBLEM STATEMENT

Emerging Stock market had been the subject of much attention in recent years with the liberalisation of their market and deregulation of their financial system. Its growing importance for its economic activity cannot be ignored. Yet stock market is susceptive to lots of volatility, high volatility is attributed to high risk, and as most investors are risk averse, they tend to shy off from the stock market due to uncertainty in expected return. (Olwendy and Omondi, 2011). In addition, It has also been identified that emerging stock market account partially for only a small segment of the global capital markets. In that way, it can be argued that stock return variation in emerging markets are more prone to local risk factor than world risk factors

With regard to the Mauritius economy, little work has been done on the dynamic relationships between stock market and macroeconomic variables. The Stock Exchange Market of Mauritius has been in operation for the past 24 years and had gained tremendous international acclamation in recent years. With changes in its policy and liberalisation of the market have been attracting a large number of foreign institution. Understanding the macroeconomic variables that could impact on the stock returns can be useful for investors, policy makers and other stakeholders as fluctuation in Stock prices tend to have repercussive effect on the not only in the portfolio performance of investors but also on the whole economy.

1.3 OBJECTIVE OF STUDY

In view of the above problem statement, the study is to contribute to the existing literature review by investigate the impact of five selected macroeconomic variables: Consumer Price Index, Exchange Rates, Money Supply, Gross Domestic Product and Treasury bills on the movement of the Mauritius Stock Market proxy by SEMDEX. This study will test for a long run and short run relationship between the stock market and macroeconomic variables using the Johansen Cointegration techniques and the VECM model. In addition, the VECM will proffer for an error correction term that would provide the speed of adjustment for the stock return examined; Granger Causality test, impulse response and variance decomposition will also be of use to observe their short term effect along with their magnitude of influence on the stock return.

1.4 STRUCTURE OF STUDY

The structure of this dissertation is divided in seven sections:

Chapter 1: This chapter provide for a general understanding of the background, problem statement and objective of this study.

Chapter 2: Observe the theoretical and empirical review of the literature on macroeconomic variables and its effect on stock return.

Chapter 3: A brief history on the selected macroeconomic variables for Mauritius and the Stock Exchange of Mauritius (SEM)

Chapter 4: Provide description of the variables and the methodology to be employed

Chapter 5: Present the empirical results and its nexus with previous research findings

Chapter 6: Discussion of the results in chapter 5, its relevance to policy implication and recommendation

Chapter 7: Offer a concluding remark on our findings, limitation of this study, and scope for further studies

CHAPTER 2

LITERATURE REVIEW

LITERATURE REVIEW

2.0 INTRODUCTION

In this chapter, theoretical and empirical review will be examined for literature. From the theoretical part, theories like the Capital Asset Pricing Model and Asset Pricing Theory will be observed; followed by empirical evidence from past studies.

2.1 THEORETICAL REVIEW

From basic financial theory, an efficient capital market is where new information are rapidly adjusted in the share price; that is the current share prices should reflect all available information. Basically, no investors should be able to employ readily available information so as to forecast future stock price movements to make a profit through the trading of shares.

Empirical researches on capital market efficiency had long been analysed, but Fama (1970) was the first to theorise capital market efficient, specially the semi-strong form efficient. As such, semi-strong form efficient implies that the stock prices should incorporate all relevant information including publicly available information; thus, this has fundamental implication for both investors and policy-markers.

For policy makers that need to implement certain macroeconomic policies, they should ensure that these policies will not conflict with stock trading activities. As for investors, the efficient market hypothesis (EMH) states that due to competition among investors whose main aim is to maximise profit, changes in macroeconomic variables will be fully incorporated in the current share prices. Consequently, no investors should be able to gain abnormal profit unless they have access to insider information- which practice is generally prohibited by laws. As such, based on the EMH, there should be no stock broking industry.

However, the concept of EMH had been overthrown based from empirical evidence accumulated throughout the past 40 years where key macroeconomics variables helped in predicting the time series of stock returns. In his articles, Economist Shiller (1981) [1] challenged the EMH by raising the question of whether the movements in stock prices can be justified by subsequent change in dividend and concludes that stock prices are too volatile to be explained by dividends and earnings.

Pricing of Assets

There exist two main theories that are most often discussed and tested which are the Capital Asset Pricing Model (CAPM) by Sharpe (1944) and Lintner (1965), and the Arbitrage Pricing Theory (APT) by Ross (1976).

2.1.1 Capital Asset Pricing Model

Initially the CAPM was a model of risk and return, but in present days this model proves to be of utmost importance in the relationship between risk and return in asset pricing. The basis of the CAPM lies in the construction of an efficient market portfolio that seeks to maximise return, given a level of risk. The expected return of an individual security is a function of its risk covariance with the market. (Kuwornu and Victor, 2011). Therefore, the requires of this model is that expected return on a stock is determine by risk free interest rate and a risk premium which is a function of the stock’s responsiveness to the overall movement in the market, that is, its beta coefficient. The formula of the CAPM is:

E (Ri) = Rf + β*E(Rm-Rf)

Where,

E (Ri): the expected return on a stock

Rf: the risk free rate

Rm: the expected market return on a portfolio

β: the risk factor

But the CAPM is often criticized as it is based on some unrealistic assumptions.

2.1.2 Arbitrage Pricing Theory

Another way of linking macroeconomic variables and stock prices is through the Arbitrage Pricing Theory (APT) developed by Ross (1976); APT is a method used to estimates the price of an asset. The theory stipulates that returns of an asset can be predicted through the relationship that exists between the same asset and other common risk factor; that is, asset’s return is dependent on various macroeconomic, market and security-specific factors. The APT is often viewed as an alternative to the CAPM, since it is argued that the assumptions governing the APT are more flexible than that under the CAPM. Both asset pricing models operate under an efficient market, as explained by Lofthouse (2001, pp 91):

"Asset-pricing models need the Efficient Market Theory (EMT). However, the notion of an efficient market is not affected by whether any particular asset-pricing theory is true. If investors preferred stocks with a high unsystematic risk that would be fine: as long as all information was immediately reflected in prices, the EMT theory would be true."

Some assumptions governing the APT:

The pricing theory assumes that asset/portfolio returns can be described by a multi-factor model and proceeds to derive the expected returns relation that follows from that assumption.

Since the intention is to maximize returns, the investor holds a number of securities so that unsystematic risk becomes negligible.

In time, arbitrageurs will exhaust all potential opportunities for riskless profits and the market will be in equilibrium.

Under the APT, investors can make profit by taking advantage of asset mispricing. While in CAPM formula requires only market's expected return, APT uses the risk premium of a number of macro-economic factors and risky asset's expected return.

APT gives the expected return on asset i as:

E(ri) = rf + bijRP1 + b2jRP2 + b3jRP4 + ... + bnjRPn

Where:

E(rj) = the asset’s expected rate of return

rf = the risk-free rate (i.e. interest on Treasury Bonds)

bj = the sensitivity of the asset’s return to the particular factor

RP = the risk premium associated with the particular factor

The APT assumes that there are n factors that cause asset returns to deviate from its expect returns. No specific number of n is giving in theory and nor does it identified these factors. To avert arbitrage, Ross demonstrates that, an asset’s expected return must be a linear function of its sensitivity to the n factors.

2.2 EMPIRICAL REVIEW

Among the pioneers to investigate the relationship between the stock market and macroeconomic variables were Chen et al. (1986) who test for arbitrage pricing theory in the US market to examine the link between stock prices and macroeconomic theory. They provide that a long term relationship exist between spread between long and short interest rate, expected and unexpected inflation, industrial production and spread between high-and low-grade bond were significant to determine risk.

Kwou and Shin (1991) examined the Korean market on a monthly basis. They found that the stock market was more sensible to real economic and international trading activities (Exchange rates, Money supply, and trade balance and production index).

Cheung et al. (1998) focus on the long term relation among Canada, Germany, Italy, Japan and USA. Having employed Oil prices, money supply and GDP as their main endogenous variables for different period; each country found that there indeed exist a common long run relationship linking stock indices and macroeconomic variables.

Since the early 2000’s, due to the rapid growth of emerging stock markets, research shifted from the analysis of the developed to that of developing nations.

In his paper, Tangjitprom (2012) classified variables used from different studies under four main groups: variables reflecting general economic conditions, other variables related to interest rates and monetary policy, price level, and some variables related to international activities.

2.2.1 Economics Conditions

Variables that are classified in this group deal with general economic conditions that can be used as proxy for cyclical factors.

2.2.1.1 Gross Domestic Products

Prominent variables that measures economic conditions are gross domestic products (GDP) or national output. In the research undertaken by Huss (2003) which applied cointergartion to investigate the linkage between the Swiss stock market and macroeconomic variables used GDP as one of its main independent variables. Wong (2010) applied advance econometrics model – AR-EGARCH and LA-VAR model- for four macroeconomic variables effects on the Chinese’s stock market. Results indicate that no causal relationship exist between stock market and real GDP, that is no real GDP is not significant in explaining stock market volatility.

2.2.12 Industrial Production Index

But many researches instead used the Industrial Production Index (IPI) as proxy for economic growth. Maysami et al. (2004) for Singapore employ the VECM for a 7 year period from February 1995 to December 2003; their results indicate that industrial production is positively and significantly related to the stock returns. Humpe and Macmillian (2007) used Industrial production index along with three other macroeconomic variables on a monthly basis for a period of 40 years (1965-2005) in VEC model. Their analysis was to investigate the long term relationship that exists for macroeconomic variables and stock prices as a comparison between US and Japan. From the cointergartion vector, stock prices were found to be positively related to industrial index for both US and Japan.

Ozbay (2009) examined the Turkish Stock market by employing Granger-causality test and conclude that industrial production is statistically insignificant meaning that the variable is neither a result variables nor the cause variable of the stock price movement.

2.2.1.2 Employment

Employment level can also be considered in examining the effects of economic conditions on stock return. The use of employment level is mostly done in event studies concerning macroeconomic news announcement. Announcement of employment level can have a more prominent impact on stock returns than the use of IPI or GDP. In the study of Flannery and Protopapadeltis (2002) employed IPI, GDP and employment announcement as determinant of stock returns and found that the IPI and GDP to be insignificant whereas employment announcement proffer a significant impact on the market. Boyd et al. (2005) also found that rising unemployment announcement can significantly affect the stock market.

2.2.2 Interest rate and Money Supply

In this group we depict variables relating to interest rates and monetary policy.

2.2.2.1 Interest rate

In general, interest rates that are employed in these studies use government securities (Three months Treasury bill and 10-years Treasury Bonds) as an alternative to interest rate.

Alam and Uddin (2009) employed time series and Panel regressions to explore the relationship between interest rates and stock prices using a mixture of fifteen developed and developing countries on a monthly basis. Results from their analysis indicate that for all countries interest rate has a significant negative relationship with stock prices. Olugbenga (2011) investigates the macroeconomic indicators on stock price in Nigeria from individual firm’s level for the period 1985 to 2009 on a quarterly basis. By employing interest rate as one of its six determinants, he found that interest rates exert negative impact on stock prices for most of the selected firms.

For Maysami et al. (2004) test for both long term and short term interest rates, results indicates that the long term interest rate exert a negative impact on the Singapore’s stock market and as for the short term interest rate wield a significant positive force on the stock market. Adam and Tweneboah (2008) for Ghana indicate the relationship between stock prices and interest rates is negative and statistically significant. As for Chen et al. (1986) indicates that interest rate had a positive impact on stock returns.

Long term interest rates (10-years Treasury Bonds) can also be employed to test for long term relationship between stock prices and macroeconomic variables, Humpe and Macmillan (2007) found that stock prices are negatively influenced by interest rates for US.

As for Gan et al. (2006) using monthly data to explore the long term and short term dynamics relationship that exist between the New Zealand Stock Market index and seven macro economic variables among which is interest rates (lending rates for long run and deposit rates for short run). By using Johansen Maximum Likelihood and Granger-causality test found that interest rate does in fact have a positive effect on the Stock Index.

The use of default spread can also be use as proxy to interest rate; default spread is the difference between the yields on risk free assets (that is, government bond) and risky assets (corporate bonds). In the paper of Chen et al. (1986), they measured the default spread by using different yield on government bonds and low-grade bonds. The results demonstrate a positive relationship between the default spread and the stock returns.

2.2.2.2 Money Supply

From previous studies, money supplies are used as monetary policies. Generally an economy is influenced by monetary policy through the transmission mechanism. Both a restrictive and an expansionary monetary policy might have bilateral effects. In case of expansionary monetary policy, the government creates excess liquidity by engaging in open market operation, which results in an increase in bond price and lower interest rates. The lower interest rate would lead to the lower required rate of return and thus, the higher stock price.

By analysing the Cypriot equity market, Tsoukalas (2003) employed the VAR model to determine the Granger causality effect for the selected macroeconomic variables on the equity market. His investigation confirms that money supply along with the other variables is strongly related to equity market. Furthermore, this result indicates that the Cypriot market is in weak form efficient, that is, where the equity price includes past information about macroeconomic policies.

Maysami et al. (2004) reveals that a positive correlation exists between changes in money supply (M2) and the Singapore’s stock returns. For the Sri Lanka stock market, Menike (2006) examines the effect of macroeconomic variables in selected companies quoted on the Colombo stock exchange market. The use of broad money (M2) results was consistent with theories, that is, M2 exert a positive impact related to stock price. Buyaksalvara (2010) results follow the trend by investigating money supply (M2) for the Turkish stock market and again a positive relationship emerge from the analysis.

Hsing (2011) used money supply (M3) as a percentage of GDP for the South African stock market by using the exponential GARCH model and result indicates that the ratio of money supply to GDP positively determine volatility in the stock market. Furthermore, Humpe and Macmillan (2007) employed M1 as a proxy for money supply and an insignificant (but positive) relationship was found between the US stock prices and Money supply.

In contrast, Rahman et al. (2009) for Malaysia investigated the linkage between the macroeconomic variables and stock prices by employing the VECM and cointegration techniques. Using monthly data for an from January 1986 to March 2008, found that Money supply along with exchange rates denote an inverse relationship with Malaysian stock market return in the long run. A bi-directional relationship was also established between the stock and the interest rates. Hussain et al. (2012) confirm the same results for Malaysia.

2.2.3 Price Level

In this section, we regroup all variables regarding price level.

2.2.3.1 Inflation

According to literature, a negative relationship is argued to exist between stock prices and inflation since an increase in the rate of inflation is accompanied by both lower expected earnings growth and higher required real returns (Ozbay, 2009). Empirical findings from Fama and Schwert (1977) and Chen et al (1986), confirm that stock returns are negatively affected by inflation. According to Chen et al (1986), they measured the inflation level through two separate variables: unexpected inflation – the difference between actual [2] and expected inflation rates, and expected inflation – the forecasted inflation reflected from other economic factors. Their results attest a negative relationship between stock returns and both of the inflation variables.

Yogaswari et al. (2012) analyse three macroeconomic variables on the Jakarta Composite Index, Agriculture sector and basic industry sector. Results conclude that changes in inflation negatively affect all three dependent variables. In the report of Humpe and Macmillan (2007) where they investigate the relationship for US and Japan; conclude that consumer price index indeed negatively affect stock prices.

Karam and Mittal (2011) for India, studied the long run relationship between the capital market and macroeconomic variables. Through the use of quarterly data and by testing for Error Correction Mechanism (ECM) indicates that inflation rates exert a negative significant impact on both the BSE Sensex and the S&P CNX. Naik and Padhi also confirm this relationship for India.

In contrast, research such as, Adam and Tweneboah (2008) for Ghana and Maysami et al. (2004) for Singapore conclude that there subsist a positive relationship between Stock prices and inflation (CPI). Moreover, Solnik and Solnik (1997) and Schotman and Schweitzer (2000) confirm fisher hypothesis that a positive relationship exist between the variables over as horizons widens.

2.2.3.2Commodity prices

Other studies have focused on oil prices and gold prices, which can be essential assets for both consumption and production specially, the oil prices. These variables can be viewed as a proxy to cost-push inflation. Olugbenga (2011) test the effect of oil prices along with five other variables in Nigeria on individual firm’s level. His analysis was made on a quarterly period from 1985 to 2009 using panel model; results perceive that for majority of the firm, movement in oil prices influence stock price movement. Hussain et al (2012) for Malaysia also demonstrate a positive relationship between oil price and Islamic financial market.

In contrast, Buyuksalvarci (2010) found a negative effect running from oil prices to stock returns for Turkey. Another alternative investment for investors is to invest in gold and some studies include gold as macroeconomic factor such as Buyuksalvarci. But the outcome from his analysis reveals that the effect of gold price is insignificant as opposed to the other macroeconomic variables used.

As for Ting et al. (2012) for the Malaysian market, they utilize crude oil price, CPI, money supply and interest rates. Oil prices were found to be insignificant in explaining movement in the Malaysian stock market under the Ordinary Least Square test and Granger causality but a negative impact was recorded in stock prices form the impulse response function. Using the Maximum Likelihood estimation, Kuwornu and Victor (2011) for Ghana found that crude oil prices do not bear any significant effect on stock returns.

Basci and Karaca (2013) in their paper explore the Turkish market using the VAR framework on monthly basis from Jaunary 1996 to October 2011 and employ Gold price along with three other variables (Export, Import and Exchange rates). It is found that Gold series induce stock prices to fall after a shock in given to gold series but in the long run it reach back the equilibrium level.

2.2.4 International Activities

With the advent of globalisation, international activities had become important since other countries fluctuation affects the local market too.

2.2.4.1 Foreign exchange rate

Exchange rates play a vital role in a country’s level of trade, which is critical to every free market economy in the world. One very basic definition of exchange rates is the rates at which one unit of a country’s currency can be exchanged into another one. As such, the observation of exchange rates is crucial and its one of the most watched, analysed and fluctuate economic variables. But exchange rates do matter on a smaller scale as well: they impact the real return of an investor’s portfolio.

There exist no theoretical consensus on the existence or the direction of any relationship between stock prices and exchange rates. But instead, classical economic theory hypothesis discuss the linkage between stock prices and exchange rates trough two models: ‘flow oriented’ model (Dornbusch and Fischer, 1980 and Gavin, 1989) and the ‘stock-oriented’ model [3] (Branson, 1983 and Frankel, 1983).

The flow oriented model assumes that important determinants of exchange rates through a country’s current account and balance of trade performance. Under this model, it is assumed that exchange rate affects the valuation of a firm through its competitiveness since it affects the cost of capital borrowed from overseas and also earnings made in foreign currencies. This in-turn influences real economic variables such as real income and output subsequently affects the current and future cash flows of companies and thus their stock prices. This goes in line with what had been discussed by Dornbuscher and Fischer which states that stock prices is defined as discounted present value of a firm’s expected future cash flows; thereby any events affecting a firm’s cash flow will be projected in the firm’s stock prices. [4] The study of Ma and Koa (1990) provides foundations for further research whereby they demonstrate that the relationship between stock prices and exchange rates can be examined in the case of an import-dominant country and export-dominant country.

Olwneny and Omondi (2011) focus their research on the Nairobi stock exchange by using EGARCH and TGARCH models; found that the impact of foreign exchange rate on stock return is low but still significant. Through the use of Johensen cointegration and VECM, a positive significant connection between stock prices and exchange rates can be found in the paper of Sohail and Hussain (2011) for the Pakistan stock market. Ozbay (2009) shows that foreign transaction has a significantly positive influence on stock price.

Naik and Padni (2012) for India investigate the relationship between stock prices and exchange rates along with four other variables using the VECM and cointegration depicts that in the both short and long run exchange rates demonstrate a negative relationship with stock prices.

Olugbenga (2011), using selected firms from the Nigerian Stock market for the period 1985 to 2009 found that exchange rates exert a negative relationship with stock prices of majority of the sampled firms. Menika (2006) for Sri Lanka and Yogaswari et al. (2012) for Indonesia found an inverse linkage between exchange rates and stock returns.

One study by Adjasi et al. (2011) investigates the relationship between stock prices and exchange rates for Mauritius along with six other African countries by employing the VAR Model. Their finding stipulates that stock returns in Mauritius is reduce due to a shock induces by the Exchange rates. One important conclusion drawn from their research is that either shocks by stock market returns or exchange rates changes seems to be more protracted in Mauritius (eight months) which indicates that misalignments in the movements of exchange rates and stock markets leave longer lasting dictions in the economy of Mauritius.

2.2.4.2 Other variables

Hsing (2011) uses the U.S. and U.K stock market index as proxy of world stock prices to examine its impact of on the South African stock price index. Using the EGARCH model, results envisage that a higher U.S. stock price or a lower U.S. government bond yield would help the South African stock market.

Alshogeathri (2011) for Saudia Arabia also include an international stock price (S&P 500) along with eight other macroeconomic variables in his VAR and GARCH model. The inclusion of this variable is because Saudi Arabia is a major oil export. The S&P 500 were found to be negatively related to Saudi Stock market through the use of the Johansen Cointegration analysis.

2.3 CONCLUSION

From the above empirical reviews, different macroeconomic variables were employed depending on the purpose of study. Furthermore, diverse methods of linking macroeconomic variables with stock prices can be found: Multiple regression techniques (Chen et al., 1986 and Buyuksalvarci, 2010), long-term relationship through the use of VAR and Cointergartion techniques (Maysami et al., 2004; Humpe and Macmillian, 2007; Ozbay, 2009), and Volatility Clustering and GARCH – family model (Wong, 2010; Hsing, 2011; Olwendy and Omondi, 2011).

CHAPTER 3

OVERVIEW

OVERVIEW

3.0 INTRODUCTION

This chapter will be focused on the stock exchange of Mauritius and its main developments that took place since its existence. In addition the monetary objective of the recent years will be reviewed.

3.1 STOCK EXCHANGE OF MAURITIUS

The Stock Exchange of Mauritius Ltd (SEM) was incorporated in Mauritius on March 30, 1989 under the Stock Exchange Act 1988, as a private limited company responsible for the operation and promotion of an efficient and regulated securities market in Mauritius.

The SEM provides for three market indices: SEMDEX, SEMTRI and DEM. The SEMDEX liststhe prices of all listed shares and with each stock weighted according to its share in the total market capitalization. Since the focus is on SEMDEX, the formula for its calculation is as follows:

The development of the SEM, being a dynamic market, has dated back from the year of incorporation 1989 till recently 2012. This is summarized in Table 3.

Table 1: Major Advancement of the SEM

1989

Starts operation with only five companies and market capitalisation of USD 92 millions

1994

Lifting of exchange control; market opens to foreigner

1997

Implementation of the Central Depository System (CDS) providing a more efficient settlement and clearing of trade.

2001

Launched the SEMATs, the use of a new generation of electronic system and the end of traditional paper trading

2003

Trading in Treasury Bill becomes accessible thus creation of secondary market for the trading of government securities

2005

Grant membership to the World Federation of Exchange and thus greater compliance with the international norms and standards

2006

Setting up of the Development and Enterprise Market (DEM), designed for small and medium enterprise

New avenue for these small firms to raise capital, attain growth and enhance their overall corporate image

2010

First exchange market in Africa to trade and settle equity products in USD

Recognise by the Cayman Island Monetary Authority thereby raising the profile of the SEM on the global field and enhance its position as an attractive Listing Venue for global and Specialised funds.

2011

The recognition of the SEM by the HMRC permits UK pension schemes to hold securities listed in the SEM official market.

2012

First listed of a GBL 1 company on the SEM

Awarded for 2 consecutive years the "Most Innovative African Stock Exchange of the Year Award"

Source: SEM

After having looked at SEM’s history and advancement, it is worth to look at its evolution of the stock prices of SEMDEX. This is graphically illustrated in Figure 1.

Figure 1: Evolution of the SEMDEX

Source: SEM

The evolution of stock prices is analysed over the years from 1989 to 2011. The data used to plot the trend line are those of daily closing stock prices in Rupees. The SEMDEX started its operation with a closing price of 100points and by the end of the 30 December 2011 closing figure at 1888.38points. The highest price to date attained by SEMDEX was in 18 February 2008 at 2101.34points.

As portray by the graph, the Mauritius stock market was not shield from the global financial crisis which effect can be clearly visible, where by 9 March 2009 the lowest prices from the financial crisis, being Rs 919.83. From there on the market recover gradually reaching the same peak as prior the crisis in mid 2011.

3.2 MONETARY POLICY IN MAURITIUS

The Monetary policies are formulated in Mauritius by the Monetary Policy Committee which was created in 2009 under the head of the BoM. The BoM regulates the price stability, exchange rates and Interest rate in Mauritius.

Table 2: Major Monetary policy measures in Mauritius

Exchange Rates

Earlier focus was on "safeguarding the internal and external value of the currency"

Now the country operates under a freely floating exchange rates where government only intervene to smooth the operations

Inflation

Like any government, price stability was of main concern after its declaration of its sovereignty

During the 1990’s administering inflation was through market-based policy instruments

These policies were found to be effective since volatile in inflation rate ranging from 8% to an average of 5% during the past 10 years.

Interest Rates

Replacement of the LOMBARD rate in 2006 for the REPO rates, where the later becomes the prime monetary instrument to signal market participants of its stance.

Source: BoM

The development of Mauritius was mostly due to an admirable monetary policy implemented all throughout the years of its existence. The growth of the country is mainly due to the heavy dependence on international transactions, especially with European market which is the country’s main exporting destination and tourism.

3.3 CONCLUSION

The performance of the SEM has been remarkably noticed over the years with the rise in its share prices. Since its inception, it has undergone much advancement. No doubt the SEM will continue to have a bright future. However we cannot deny limitations associated with it.

CHAPTER 4

METHODOLOGY

METHODOLOGY

4.0 INTRODUCTION

This chapter research for the appropriate variables to be included in our model to investigate the nexus between stock prices and macroeconomic variables. Following that, theory behind the econometric techniques that to be used in the next chapter that would confirm the hypothesis relationship based from theory.

4.2 VARIABLES

The Asset Pricing Theories does not provide for precise events nor which economic factors are the most influential on the asset price. This provides for an opportunity to investigate on the relevant economic factors to be included in our model. The Discounted cash flow model provides or the present value model (PVM) offers a way in which these variables can be determined. As such the PVM states that the price of a stock is the present discounted value of the expected future dividends received by the stock holder. PVM can be expressed as follows (Alshogeatheri, 2011):

Where, Pt: Stock prices

E(Rt+i): Stock price affected by any possible changes in the expected stream of return

Kt: Discount rate of future cash flow

Factors that may, directly or indirectly, affect expected returns and will affect the stock prices; that is, relevant macroeconomic information may be analysed as long as they impact on stock prices of the discount rates, kt or both.

For this study, five macroeconomic variables will be employed that we expect to have a significant effect on Mauritius stock market (SEMDEX) based on empirical review. The analysis will be on quarterly period of 10 years from January 2001 to December 2011. The five variables are: Consumer price index (CPI), Money Supply (MS), Exchange rates (ER), Gross Domestic Product (GDP) and 3-month Treasury bill (TB).

(I) Dependent variables - SEMDEX

The SEMDEX is an all-share index of prices of all listed shares. It is a weighted index and each stock is weighted according to its shares in the total market capitalization. Thus, changes in the SEMDEX are dominated by changes in the prices of shares with relatively higher market capitalization.

(II) Consumer Price Index (CPI)

The Consumer Price Index is used as proxy for inflation rate in this study. The choice of CPI as inflation rate is because its calculation is based upon average change in prices of goods and services. In addition, inflation rate is normally translated into the nominal interest rates. An increase in nominal interest rates increases discount rate which results in reduction of present value of cash flows.

(Sohail and Hussian, 2011; Buyuksalvarci, 2010)

Hypothesis: Inverse relationship

(III) Treasury Bill (TB)

Most studies utilize the three-month Treasury bill as proxy for interest rates. The intuition regarding the relationship between interest rates and stock prices is well established, suggesting that an increase in interest rates increases the opportunity cost of holding money and thus leading to substitution between stocks and interest bearing securities and hence reduction in stock prices.

(Naik and Padhi, 2012; Menike, 2006)

Hypothesis: Inverse relationship

(IV) Exchange rates (ER)

From past studies, results had found that exchange rate and stock prices are closely related. Local currency vis-a-vis the US figures will be more appropriate for our study since Mauritius is a major import oriented country. An appreciation in the home currency will render exported goods more expensive on the international level, thereby leading to fall in sales which in turn cause stock prices to fall.

(Kuwornu and Victor, 2011; Ibrahim and Aziz, 2003)

Hypothesis: Inverse relationship

(V) Gross Domestic Product (GDP)

GDP is used as proxy to measure the overall performance of an economy. It is the measure of all currently produced final goods and services valued at market prices and is thus an aggregated value of all the industries in an economy. Investor will survey this figure to assess their future investment decision; in a period of economic downfall, investors' confidence on the expected return on stock prices will lessen, thereby reducing their investment on stock market which consequently leads to a fall in the stock prices.

(Gan et al., 2006; Olugbenga, 2011)

Hypothesis: Positive relationship

(VI) Money supply (MS)

Broad money (M2) is used as a proxy of money supply. Increase in money supply leads to increase in liquidity that ultimately results in upward movement of stock prices.

(Hancocks, 2010; Rahman et al., 2009)

Hypothesis: Positive relationship

The variables employed in this study do not, by any means, enclose all factors that affect stock price movement instead these variables are normally considered important factors that can cause fluctuations in stock return.

4.2.7 Data Collection:

Secondary time series date were collected and used in the analytical part.

Table 3: Source of Data

Variables

Proxy

Explanation

Units

Source of data

Stock Market Return

SEMDEX

market index of Mauritius

30 March 1989 = 100

SEM website

Consumer price index

CPI

Consumer price index as proxy for inflation

Index Number

CSO

Money Supply

MS

Broad Money (M2)

Rs Million

BoM website

Treasury Bill

TB

3-month Treasury Bill

Yield

BoM Website

GDP

GDP

Gross Domestic Product at Market Price

Rs Million

CSO website

Exchange Rate

ER

MRU vis-a-vis US Dollar

Rs per US$

BoM website

Econometrics tools employed: Eviews 6 and STATA 11

4.3 ECONOMETRIC TECHNIQUES

This study compiles the methodology employed by authors such as Adam and Tweneboah (2008), Gan et al. (2006), Sohail and Hussain (2011) and Rahman et al. (2009). Form their analytical part they utilized Cointegration and restricted VAR model to investigate the relationship linking stock prices with their selected macroeconomic variables.

4.3.1The Model:

Based on both theoretical and empirical reviewed, a multivariate process with the hypothesis that Stock Price along with its independent variables.

Economic function:

SEMDEX t= f (ERt, INTt, GDPt, MSt, CPIt) [1]

The variables were then converted in natural logarithm so as to reduce the gap of the data between them.

LSEMDEXt = β0 + β1 LERt+ β2 LTBt + β3 LGDPt + β4 LMSt + β5 LCPIt + ut [3]

Where β0 is the constant

β is coefficient of the variables

LSEMDEX: Natural logarithm of stock market return in Mauritius at t year

LER: Natural logarithm of exchange rates in Mauritius at t year

LTB: Natural logarithm of Treasury bill in Mauritius at t year

LMS: Natural logarithm of Money Supply (M2) at t year

LCPI: Natural logarithm of Consumer price index at t year

LGDP: Natural logarithm of Export at t year

ut is the error of the regression

4.3.2 Stationarity Test

For any time series analysis, the first step required is to determine whether the data being tested is stationary or not. Data series is said to be stationary, if its mean variance is constant (that is, no changing with regards to time) over time and covariance between two time periods depends only on the distance or gap between the two time periods and not on actual time at which covariance is computed. Therefore prior to performing any test, it is advisable to perform tests for stationarity using the Augmented Dickey Fuller (ADF) and the Philips-Perron (PP). Regression of non-stationary time series variables may often give spurious or nonsensical results.

4.3.3 Co-integration (Long Run Relations)

Many economic time series data such as stock prices, dividends and other macroeconomic variables share long-run relationship. As such, cointegration examines long run relationships that exist between a set of variables, that is, the tendency that variables move together in the long run.

Two methods exist to test for long-run relationship in literature, the Engle and Granger (1987) cointegration test and the Johansen-Juselius (1990) cointegration procedures. The former is appropriate for bivariate analysis, while the later is suitable when there are more than two variables to be analysed.

Since the purpose of this study is to investigate the short and long term relationships among stock prices and five macroeconomic variables, we will focus only on the Johansen-Juselius (JJ) (1990) cointegration test.

The JJ test is a statistical method to test for whether there is cointegration or not among variables. This method is based on a Vector Auto Regression (VAR) model to establish any long-run relationship that may exist among the analysed variables. Under the JJ test, we assume that all the variables are endogenous which is not the case in Engle-Granger (1987) [5] . The presence of a co integration relation(s) forms the basis of the vector error correction model (VECM) specification. Identifying the number of co integration vectors within the VAR model is the basis for this procedure. (Vongrai, 2010).

To indentify the number of cointegrating vectors, a likelihood ratio test of hypotheses procedure is used. Prior to performing the Johansen procedures a series of steps need to be considered:

Lag selection

A lag length test is used to find the number of lag value to be included on the model. Choosing the appropriate lag is important since using too much lags tend to reduce the power of the test and a loss of degrees of freedom; as for choosing too few lag, this may prevent the test to capture the dynamics of the actual error correction process and its standard errors. Two different information criteria can be employed: Akaike Information Criterion (AIC) or the Schwarz-Bayesian Criterion (SBC). But different studies that employed cointegration analysis have used both AIC and SBIC with neither alternative firmly agreed upon between studies.

Rank determination

Two ways to perform the likehood ratio test was proposed by Johansen:

Trace test - the null hypothesis is that the number of cointegrating vectors is less than or equal to r, against a general alternative hypothesis that there are more than r.

Maximal Eigenvalue test - The null hypothesis is that there are r cointegrating vectors present against the alternative that there are (r + 1) present.

Both tests can be used to determine the number of cointegrating vectors present, although the number of cointegrating vector may differ.

4.3.4Vector Error Correction Model (VECM)

The VECM explore the long run causality and the short term dynamics if there is evidence of cointegration among the variables. The VECM is a restricted VAR designed for use with non-stationary data that was found to be co integrated. The VECM is opted over the VAR model as the former dispense for a far better understanding of the nature of non-stationarity among different compound series and it also improve longer term forecasting over an unconstrained mode. Besides, the introduction of a VECM helped in correcting a disequilibrium that may shock the whole system.

General form of a VECM is as follows:

Δ Yt = λ1 Δ Yt-1 + λ2 Δ Yt-2 + …… + λ p-1 Δ Yt-(p-1) + Π Y t-1 + Ut

Where Y is the matrix consistent of number of variables in row by number of cointegrating vectors in column and Π is the dynamic adjustment.

The dynamic adjustment (also known as the error correction term) is the speed at which short run converge toward an equilibrium in the long run.

4.3.5Granger causality Test

Granger Causality is used to test on short term relationship between endogenous and exogenous variables. This test favoured stationary data over non stationary data. For instance, the Wald statistics better suited for VECM since it find the causal effect on the dependent variable. This test can be used to confirm results obtained from the VECM for the short run.

4.3.6Variance Decomposition (VD)

Normally for any variables, short run variations are due to their own shocks, but over times other variables contribute to these changes as well. The variance decomposition or the forecast error variance decomposition examine this phenomenon, that is, which macroeconomic factors explain a significant part of the variation in stock prices on the short and medium run. The VD is constructed from a VECM model. It can directly address the contribution of macroeconomic variable in forecasting the variance of stock price (Kazi, 2008).

4.3.7 Impulse Response Function (IRF)

The IRF is an important tool to interpret VAR results. It describes how economy reacts over time to exogenous impulse know as ‘shock’. IRF test the reaction of endogenous variables at the time of a shock and over subsequent points in time. Moreover, the IRF are reliable only with a stationary time series the data has been turned into stationary after the second difference. It acts as an econometric technique which has been employed to investigate the short-run impact caused by the VECM when it received some impulses.

4.4 CONCLUSION

As series of test would be carried out by using the five macroeconomic variables to investigate the different relationship with that exist with stock return. Tests such as: Unit root test, Johansen Cointegration tests, Granger Causality test, Variance Decomposition and Impulse Response Function. Each of these tests has different purpose thereby yielding different results in modelling the relationship between these variables and stock return.

CHAPTER 5

ANALYSIS

ANALYSIS

5.0 INTRODUCTION

This chapter will provide the results from the suggested methodology in the previous chapter. In the first instance stationary test based on the ADF and PP will be performed followed by Johansen Cointegration, VECM, Granger Causality, Variance Decomposition and Impulse Response Function. An explanation will be discussed based on the findings that are presented in table form.

5.1 STATIONARY TEST

The first step prior to the performance of the Johansen cointegration test is to determine the order of intergration of our variables. Two tests will be conducted to test for stationarity simultaneously: the ADF and the PP test so as to ensure the reliability of our results.

H0: the variables contain Unit roots

H1: the variables does not contain Unit roots

Appendix 1 suggests that no variables are stationary whether under the ADF or PP test statistics. That is, the critical value is found to be greater than the test statistics at the 5% level of significance; the null hypothesis of the unit roots for all variables cannot be rejected, resulting in the acceptance of the null hypothesis that all variables have unit roots. To determine whether the variables are stationary, we perform these tests using the first difference of the variables.

From appendix 2, after computing the test using first-difference data, both tests converge toward the rejection of the null hypothesis, that there is no unit root. We accept the alternative hypothesis that all variables become stationary in their first differences; we conclude that all of the series are of order of integration of I(1). Accordingly the performance of the co integration properties is the next step.

5.2 COINTEGRATION ANALYSIS

The cointegration analysis is conducted using the Johansen-Julius procedures to depict for any long run equilibrium between the stock return and the macroeconomic variables. This process will thus enable us to formulate an error correction model if cointegration is found. Before conducting the Johansen procedures, two pre-condition must be satisfied: all variables should be integrated of same order and the secondly, the linear combinations of the variables from the regression of the non-stationary variables which according to Brooks (2008) ensure that we eliminate spurious relation and therefore will share common stochastic trends. Since figures from appendix 2 confirms these pre-requirements we can then proceed with our analysis.

5.2.1 Lag Length Selection

The VAR model is sensitive to the lag selection, thus for choosing the appropriate lag order for our VAR model, the information criteria approach such as the Schwarz Bayesian Criteria (SBC) and Akaike information criteria (AIC) test will be used.

Lag 1 was selected for this analysis based (Appendix 3) on the Schwarz Bayesian Criteria (SBC) and following the Sohail and Hussain (2011) and Naika and Padhi (2012) steps. We reject the AIC since we have a restraint data set and also not to lose too much degrees of freedom.

5.2.2 Johansen Cointegration test

The next step is finding the number of cointergrating vectors that our model contain. The number of long run vectors can be obtained using the Trace statistics and the Max Eigen Statistics.

From Appendix 3 it can be observed that both the Trace statistic and Maximum Eigen statistic conclude that there exists at least one cointegrating relationship [6] . Therefore, a long run relationship exist between the Stock Return and our selected macroeconomic variables in Mauritius and the finding is consistent with a large body of empirical studies including Adam and Tweneboah (2008), Sohail and Hussain (2011), Maysami et al. (2004), Humpe and Maclillian (2009) and Gan et al. (2006) among others.

5.3 VECTOR ERROR CORRECTION MODEL

Given the existence of at least one cointegrating vector among the variables, we estimate a VECM including the error correction term to investigate the dynamic of our model.

5.3.1 Long Run Estimates

The analysis provides a normalized cointegrating vector for the stock return.

Table 4: Normalise Cointegration Coefficient

Source: Computed [*** Variables significant at 1 percent significance level]

The normalized cointegrating coefficients for our long run equation can be written as such:

LSEMDEXt= – 7.494LMSt -1.479LCPIt + 11.01LGDPt - 0.725LTBt - 9.540LERt – 20.02740

In accordance with our hypothesis, the results demonstrate a negative interaction between consumer price index (inflation) and stock return (LSEMDEX) but contrary to expectation it’s also found to be insignificant. Buyukalvarci (2010) for Turkey also found no significant relationship between stock return and inflation. A rational explanation for such results is that the stock market predetermines the inflation figures nearly accurately prior the announcement of actual figures. Therefore, this is in line with the objective of the government in pricing stability. Negative yet significant relationship between CPI and stock prices can be also be observed in Chen, Roll and Ross (1986), Yogaswari et al. (2012) and Ting et al. (2012).

From the equation, a 1% increase in Treasury Bills (LTB) leads to a fall of 0.42% in LSEMDEX. The negative relationship found in the Treasury bill is in accordance with results from Olugbenga (2011), Adam and Tweneboah (2008), and Kuwon et al. (2011). The reason press forward by various authors for this relationship is that Treasury bill can be a better proxy for nominal risk-free component used in the discount rate in the stock valuation models and can also be used as alternative investment tools for the Mauritian investors. A high rate in Treasury bills would cause rational investors to shift from risky asset with high return to investment in Treasury bills. But the sales of Treasury bills is restricted to only the residents of Mauritius thereby restraining its power as an alternative investment tools in Mauritius as we should not forget that a strong proportion of transactions done on the SEM are by foreigners.

Contrary to our expect hypothesis, Money supply and stock prices project a negative relationship. A 1% rise in Money supply will shrink stock prices by 7.49 percent. Alshogeathri (2011) for Saudi Arabia, Rahman et al. (2009) and Hussin et al. (2012) for Malaysia, and Sohail and Hussain (2011) for Pakistan found similar negative relationship between money supply and stock prices. This negative relationship may be due to an excessive money supply that would lead to a higher inflation rate causing an increasing in the discounting rate which in turn will lead to a fall in share prices. (Gan et al., 2006)

Stock return and exchange rates demonstrate a negative relationship whereby a 1% increase in LER induces LSEMDEX to fall by 9.54%. This negative and highly significant relationship denotes that the economy is highly dependent on international trade. Common results were found for Singapore (Maysami and Koh, 2000) and Malaysia (Ibrahim and Aziz, 2003). As the local’s currency depreciates against the U.S. dollar, imported products become more expensive. As a result, if the demand for these goods is elastic, the volume of imports would increase, which in turn causes lower cash flows, profits and the stock price of the domestic companies. While currency depreciation encourages exports, it increases cost of production through increasing domestic prices of imported capital and intermediate goods.

The 1% rise in GDP Lead to an 11.01% increase in LSEMDEX, therefore stock return is positively related to economic activity as proxies by the GDP. This result is consistent with our expectation and a large amount of empirical studies (Sohail and Hussain, 2011; Humpe and Macmillian, 2009; Fama, 1981; Naik and Padni, 2012). During periods of high economic growth, and a high level of confidence will prevail in the economy, investors will expect a greater return and thereby will be willing to invest on the stock market, leading to a rise in stock prices.

All variables employed to test for the Long run estimators conclude to be significant except for CPI in explaining the long term variation in stock prices.

5.3.2 Short Run Estimates

To uncover the short run relationships among the variables, a vector error correction mechanism (VECM) is employed. The error correction term (ECT) as per its definition should be negative and significant and results from our VECM for LSEMDEX demonstrate that both the sign and magnitude of the error correction coefficient is in accordance with the theory. Therefore it clearly demonstrates the direction and speed of adjustment towards the long-run equilibrium path.

Table 5: Vector Error Correction Estimates

Source: Computed [*, **, and *** indicate significance at 10%, 5% and 1% respectively]

While examining the short rum estimates, the equation having LSEMDEX as the dependent variables will be used; we discern that only past LGDP, past LER and past LSEMDEX are statistically significant in explaining the short run effect on LSEMDEX. Only past LGDP create a negative impact on the current moves in stock prices, as for past LSEMDEX and LER they yield a positive movement in current stock prices. Therefore, an increase in past stock prices of 1 percent will induce a 0.28 percent increase in the present LSEMDEX in the short run.

As for past LGDP, a 1percent increase will cause LSEMDEX by 0.47 percent to decrease in the short run. Acikalin et al. 2008 for Malaysia also denote the same relationship between stock prices and past GDP.

We now examine the past moves of LSEMDEX in creating impact on current changes in macroeconomic variables. It shows that past moves of LSEMDEX lead to positive changes in current LMS, LGDP and LTB. Form this we can deduce a bidirectional relationship between stock prices and GDP.

As indicates in table 5, the estimated coefficient from the ECT (-1) is -0.098 suggest that, in the absence of variation in other variables, the LSEMDEX will correct itself from the deviation in the long run by only 9.6 percent per quarter. Therefore, it will take the LSEMDEX approximately 11 quarters to correct itself (1/0.098=10.2).

Nevertheless, the R2 (0.668780) implies that 66.8 percent of the variation of the LSEMDEX is accounted by Interest rates, Exchange rates, Money Supply, Export and Inflation and thus the overall goodness of fit of the VECM is satisfied.

The VECM results confirm the existence of a definite causality between stock prices and its macroeconomic variables in the short run. But for the sake of consistency, though the VECM analysis confirm for relationships between the variables and to provide for a more robust conclusion, causality test will be performed so as to back up results from the VECM estimation.

5.4 GRANGER CAUSALITY TEST

A short run causality test (Ward test) through the use of the Chi-square will be performed on the VECM. The null hypothesis that no causal direction exist will be tested against the alternative hypothesis that a direction of causality exist among the variables.

Table 6: Granger Causality Test

Source: Computed

Table 6 demonstrates that out of the five independent variables three have a granger-causal effect on stock return (LSEMDEX). From empirical results above, CPI and Money supply does not have any effect on the stock prices in the short run. This is in contradiction to various previous studies (Rahman et al. (2009) and Ting et al. (2012)) where they found that all their variables employed have a causal interaction with stock price but not vice versa.

The causal relation from exchange rates to stock prices supports the Flow-Oriented Model (Dornbusch and Fisher, 1980). Therefore, depreciation in the local currency makes local firms more competitive on the international market this leads to a higher export which in turn increases income and thus the firm’s stock prices. Tsoukalas (2003) for Cyprus, Ozbay (2009) for Turkey and Naik and Padni (2012) for India concord with this result.

For both LGDP and LTB a bidirectional causal relationship exist with stock prices. The results concord with the findings of Rahman et al. (2009) for Malaysia, where a bi-directional relationship was found between interest rates and stock prices. The causal relationship that run from stock prices to GDP and vice versa support what have been proposed by the VECM estimation. This relationship can be explained that the ratio of capitalization of the stock to GDP as compared to other market is quiet high. The results tally with Basci and Karaca (2013) for the Turkish market and Acikalin et al. (2008) where both found a bidirectional causal relationship between stock prices and GDP.

In a nutshell only exchange rates, GDP and Treasury bill are found to constitute to an important part in the Granger causal effect of stock price in the short run.

5.5 DYNAMIC ANALYSIS

Although we performed the causality test, the later by definition, does not give neither the strength nor does it describe the relationship between the variables over time. Therefore, to examine the response of Mauritius stock returns to shocks of macroeconomic variables both the impulse response function and the Variance decomposition (VD) are used to estimate the responses. For this section, we will generate the VD and IRF to examine the short-run dynamic interactions among the variables.

5.5.1 Variance Decomposition

As per Brooks (2008), he states that the variance decomposition determines how much of the forecast error variance of each of the variable can be explained by independent shocks to the other variables. Therefore, the test determines the proportion of the movement in the dependent variables (LSEMDEX) that is due to their ‘own’ shocks and versus shocks of the other variables (Brooks, 2008).

Table 7: Forecast Error Variance Decomposition

Source: Computed

From the table 7, results demonstrate that all of the variables explain a substantial amount of variation on stock return over both the short and medium run.

In the first period 100 percent of SEMDEX’s variance can be explained by its own innovation but gradually dropped to 56.54 percent by the end of the period. Exchange rate is the only variable with the smaller impact of 1.19 percent on stock return in the second quarter but by the end of the analysed period it is also the variable which causes the greatest variations of 21.55 percent in LSEMDEX.

It can also be noted that from period to period, the impact of the independent variables becomes more prominent on the dependent variable (LSEMDEX) except for LCPI which show a fall from 2.52 percent in the first period to that of a 0.66 percent by the end.

During the first quarters, the overall impact of the independent variables is smaller in the short run but exert bigger force on the dependent variable in the medium term.

5.5.2 Impulse Response Function

The IRF tracks the response of a variable over time after a shock to the VAR system. The persistence of a shock indicates how quickly the system returns to equilibrium. (Kuwornu and Victor, 2011). Also, the IRF allows us to determine the magnitude, direction and length of time that the LSEMDEX is affected by a shock of a variable in the system, holding all other variables constant.

A shock in stock prices induces a rise in its own stock prices until the 3rd period. From there on a small decline in stock is observed until the 5th quarter where it remains constant throug