THE SOUTH AFRICAN STOCK MARKET AND THE ECONOMIC ENVIRONMENT

Get Complete Project Material File(s) Now! »

A REVIEW ON EXISTING STOCK MARKET MODELS

INTRODUCTION

Studies modeling stock markets can be divided into two broad categories, namely those that test stock market efficiency and those that model stock prices or stock returns. Studies modeling stock market efficiency are basically evaluating the efficient market hypothesis and the random walk model. On the other hand, studies that model stock market prices or returns are based on the theory of the present value model. Within the latter category, studies can be classified as either structural models that try to develop and estimate a model of the stock market, or studies that evaluates the relationship between stock market and macroeconomic variables.
In this study a structural model of the South African stock market will be estimated and therefore the focus of the literature review will be on literature estimating structural stock market models rather than the literature evaluating stock market efficiency. However, stock market efficiency has important implications for the profitability of a trading rule based on technical analysis versus trading based on a structural model. Therefore, although the focus of this study and hence the focus of the literature review will be on the structural models of the stock market, it is important to also include a brief overview on the literature evaluating stock market efficiency.
In this chapter, a brief overview of studies modeling stock markets will be given. First, the literature on international stock market models will be reviewed, differentiating between studies evaluating stock market efficiency and those that estimate structural models of the stock market or analyze the relationship between the stock market and specific macroeconomic variables. The latter category will distinguish between studies that modeled stock prices or those that modeled stock returns. This is followed by an overview of the literature on empirical studies of the South African stock market, again distinguishing between studies evaluating the efficiency of the JSE and studies modeling stock prices or stock returns.

INTERNATIONAL STUDIES

Studies Evaluating Stock Market Efficiency

Stock market efficiency has fundamental implications for stock market analysis and trading. If stock markets are not efficient, stock prices are forecastable from past price behavior alone. The random walk theory, which assumes that consecutive price changes are independent and identically distributed over time, is central to the testing of the ability of past returns to predict future returns (Thompson and Ward 1995). If prices follow a random walk, it means that yesterday’s price change should not be related to the price change of today, or any other day since it should be independent. The implication for trading is that future price movements cannot be predicted successfully on the basis of historic price movements.
Empirical studies have mainly used three econometric techniques to evaluate stock market efficiency, namely serial correlation tests, the runs test and the variance ratio test. Tests for weak form efficiency can be divided into two broad categories. The first group includes studies that test whether trading rules based on exploiting possible systematic patterns in share prices can yield abnormal profit, in other words can beat a random selection of securities. Spectral analysis, serial correlation tests and the runs test are usually utilized to analyze the dependence of share prices. Although share prices are seldom perfectly independent, stock market investors are mostly concerned with whether the dependence is sufficient to allow the history of the series of price changes to be used to predict the future in such a way that the expected returns would be greater than under a simple buy-and-hold model (Thompson and Ward 1995).
The second group of weak form efficiency tests comprises studies testing the statistical dependence in changes in share prices, attempting to determine whether share prices have sufficient dependence to make it possible for investors to predict future share prices by studying past trends. Weak form efficiency is invalidated if a trading rule, in other words a strategy for buying and selling securities based on objective signals, consistently outperforms a simple buy-and-hold portfolio with equivalent risk (Thompson and Ward 1995).
Tests of semi-strong form market efficiency generally evaluate the speed and accuracy of market adjustment to specific new information that affects the intrinsic value of the security. These studies test whether the market moved in the right direction and the speed of market adjustment following a specific type of information generating event. Information generating events include earnings announcements, changes to dividend policy, capitalizations, large secondary offerings of common stock, changes in the discount rate and changes to accounting methods. The main methodology followed in these studies is to compare expected share prices to actual share price performance, where the expected share prices are usually determined with some form of asset pricing model. The residuals are then analyzed to determine the impact of the information on share prices and whether share prices adjusted rapidly and accurately to this information (Thompson and Ward 1995).
Tests of strong form market efficiency entail evaluating whether specific investors or groups of investors have monopolistic access to non-public information relating to price formation. The rates of return on portfolios of investors that have access to private information, usually professional portfolio managers of unit trusts and investment funds, are compared to that of a passive buy-and-hold-the-market strategy. If such an investor consistently and significantly outperforms the market, it indicates either exceptional skills or access to special information, which negates the strong form of the efficient market hypothesis (Thompson and Ward 1995).
The empirical evidence on market efficiency in international stock markets has been inconclusive. While many studies found that markets are efficient (see e.g. Kavussanos and Dockery 2001; Chen, Kwok and Rui 2001; Nieto, Fernandex and Munoz 1998), there are also studies that found evidence against market efficiency (see e.g. Omet, Khasawneh and Khasawneh 2002; Siourounis 2002; Hasan, Samarakoon and Hasan 2000; Mecagni and Sourial 1999).

Structural Stock Market Models

Stock price models

The literature on modeling stock market prices instead of modeling stock market returns (i.e. changes in the stock market prices) is quite sparse. Studies that did model stock prices all employed cointegration techniques and used the present value model as theoretical basis. Even though most of these studies used the Gordon-Shapiro (i.e. constant growth) version of the present value model, different studies interpreted the empirical implications of this model differently.
According to the present value model, stock prices are a function of future dividends, the discount rate and the growth rate. In empirical studies, dividends are often proxied by variables such as economic growth or industrial production, while the discount rate is specified as the long-run interest rate to which a risk premium is sometimes added (see section 3.3.1).
Harasty and Roulet (2000) used cointegration techniques to model the stock prices of 17 developed countries. They argue that economic theory can explain the long-run trend of the stock market, but that short-run movements are driven by variables other than those dictated by theory and hence it can only be determined empirically. Therefore, they estimate the long-run behavior of stock prices based on the present value model and then empirically try to explain the fluctuations of the market around this long-run trend. Using the Engle-Granger cointegration technique, they showed that stock prices are cointegrated with earnings (a proxy for dividends) and the long- term interest rate in each country (except the Italian market for which the short-term interest rate was used). The main variables that explained the short-term fluctuations were short-term interest rates, exchange rates and the spreads between domestic long- term and short-term interest rates, as well as between domestic and foreign interest rates.
Following a similar approach to model the long-run behavior of Spanish stock prices, Ansotegui and Esteban (2002) also based their model on the present value model. They showed that stock prices cointegrate with industrial production (used as proxy for dividends), inflation and the interest rate. Han (1996) interpreted the present value model differently and tested for cointegration between stock prices and dividends of the Standard and Poor stock index. He found that neither the levels nor the logarithmic transformations of stock prices and dividends are cointegrated and therefore concluded that the present value model doesn’t hold for the Standard and Poor stock index. However, Yuhn (1996) argues that the present value model doesn’t imply cointegration between stock prices and dividends. By using extensive mathematical derivations, he shows that the present value model rather implies that the sum of current stock prices and dividends should cointegrate with lagged stock prices. When he tests the present value model with this specification, he found little evidence supporting linear cointegration but overwhelming evidence of non-linear cointegration.
There is evidence that the present value model has been interpreted in various ways in the literature. This resulted in different model specifications in different studies, which has a crucial impact on their results especially in terms of whether they reject validity of the present value model. In addition to the model specification differences, different authors have used different proxies for dividends and discount rates.

Stock return models

 The studies that have modeled actual, expected or excess stock market returns can be divided into two categories. The first category includes studies that test whether stock markets are efficient, while studies in the second category analyze the relationship between the stock market and macroeconomic variables. Studies in the latter category either evaluate the bivariate relationship between stock prices and a macroeconomic variable, or try to build a model for stock prices
As set out in chapter three, the present value model asserts that stock prices are determined by dividends and the discount rate and are hence influenced by macroeconomic variables that influences or proxies dividends or the discount rate. It follows trivially that the systematic forces that influence stock prices and hence returns, are those that influence the discount factor or dividends. Since the seminal article by Chen et al (1986), the influence of variables such as interest rates and inflation on the discount rate and of the economic growth on dividends has been well established. However, different studies have defined the discount rate differently and also used different proxies for economic growth and dividends.
The relationship between stock prices and interest rates has received considerable attention in the literature. A distinction has to be made between the influence of the long-term and the short-term interest rates, since the rationale for their relationships with the stock market differs. The proxy hypothesis of Fama (1981) argues that expected inflation is negatively correlated with anticipated real activity, which in turn is positively related to returns on the stock market. Therefore, stock market returns should be negatively correlated with expected inflation, which is often proxied by the short-term interest rate. On the other hand, the influence of the long-term interest rate on stock prices stems directly from the present value model through the influence of the long-term interest rate on the discount rate (see section 3.3).
Lee (1997) used three-year rolling regressions to analyze the relationship between the stock market and the short-term interest rate. He tried to forecast excess returns (i.e. the differential between stock market returns and the risk-free short-run interest rate) on the Standard and Poor 500 (S&P500) index with the short-term interest rate, but found that the relationship is not stable over time. It gradually changes from a significantly negative to no relationship, or even a positive although insignificant relationship.
Zhou (1996) also studied the relationship between interest rates and stock prices using regression analysis. He found that interest rates have an important impact on stock returns, especially on long horizons, but the hypothesis that expected stock returns move one-for-one with ex ante interest rates is rejected. In addition, his results show that long-term interest rates explain a major part of the variation in price-dividend ratios and suggests that the high volatility of the stock market is related to the high volatility of long-term bond yields and may be accounted for by changing forecasts of discount rates.
Rather than using either short-term or long-term interest rates, Campbell (1987) analyzed the relationship between the yield spread and stock market returns. He argues that the same variables that have been used to predict excess returns in the term structure also predicts excess stock returns, deducing that a simultaneous analysis of the returns on bills, bonds and stock should be beneficial. His results support the effectiveness of the term structure of interest rates in predicting excess returns on the US stock market.
Kaul (1990) studied the relationship between expected inflation and the stock market, which, according to the proxy hypothesis of Fama (1981) should be negatively related since expected inflation is negatively correlated with anticipated real activity, which in turn is positively related to returns on the stock market. Instead of using the short- term interest rate as a proxy for expected inflation (like for example Lee (1997)), Kaul (1990) explicitly models the relationship between expected inflation and stock market returns. His results is supportive of Fama’s (1981) proxy hypothesis and showed that the relationship between stock returns and expected inflation in the US is significant and negative.
Spyrou (2001) also studied the relationship between inflation and stock returns but for the emerging economy of Greece. Consistent with Kaul’s results, Spyrou (2001) found that inflation and stock returns are negatively related, but only up to 1995 after which the relationship became insignificant. He ascribes the change in the relationship to the increased role of monetary fluctuations, in line with the argument of Marshall (1992) that the negative relationship between stock returns will be less pronounced during periods when inflation is generated by monetary fluctuations.
In addition to inflation and interest rates, Leung, Daouk and Chen (2000) included the lagged stock market index and economic growth as explanatory variables in their stock market models for the US, UK and Japan. They model not only the stock market index, but also turning points in the stock market index in order to compare the profitability of trading rules based on the two approaches. To model stock prices they employ adaptive exponential smoothing techniques, the VAR-Kalman Filter, a transfer function and neural networks. They model turning points in the stock market with linear discriminant analysis, a logit model and neural networks. Their results suggest that classification models outperform level estimation models in terms of predicting the direction of the stock market movement and maximizing returns.
Fang (2002) argued that exchange rates could also influence stock prices. This should especially be relevant in the current globalized economy. His results confirmed that currency depreciation adversely affects stock returns and increases market volatility over the period of the Asian crises (1997-1999). The implication for investors is that they have to evaluate the stability of foreign exchange markets prior to investing in stock markets. However, this study only covered crisis periods and the results might differ for normal periods.
Black and Fraser (1995) argue that the predictable variation in excess stock returns is a rational response to the general level of expected business conditions. Following the present value model, stock prices are in part determined by future dividends, which in turn are influenced by the future state of the economy. Since current financial variables reflect the expected future state of the economy, it should be able to predict the conditional risk component of excess returns. The results of their Garch-M model are supportive of their hypothesis that financial variables, specifically the term spread, default spread and dividend yields, influence UK stock returns.
Chen (1991) follows a similar line of reasoning than Black and Fraser (1995). He argues that stock market returns are a function of expected economic growth through its influence on dividends and economic growth in turn is a function of so-called “state variables” such as interest rates, interest rate spreads and dividend yields. In addition, the uncertainty regarding future economic growth (or dividends) also plays a role in determining the stock prices, so his stock market model also includes the volatility of economic growth as explanatory variable. He empirically showed that lagged economic growth, the default spread, the term spread, short-term interest rates and the dividend-price ratio are important determinants of future stock market returns in the US. In addition, expected excess market return is negatively related to recent economic growth and positively related to future growth.

LIST OF TABLES 
LIST OF FIGURES 
1. INTRODUCTION AND BACKGROUND
1.1 Introduction
.2 Objectives and methodology
1.3 Contributions of this study
1.4 Outline of the study
2. THE SOUTH AFRICAN STOCK MARKET AND THE ECONOMIC ENVIRONMENT
2.1 Introduction
2.2 The structure of the Johannesburg stock exchange
2.3 The role and functioning of the South African financial market and the Johannesburg stock exchange
2.3.1 The role and functioning of the South African financial market
2.3.2 The role and functioning of the Johannesburg stock exchange
2.4 The socio-economic environment
2.5 The institutional and policy setting
2.6 The impact of globalization and South Africa’s emerging market status on the JSE
2.6.1 Globalization and global financial revolution
2.6.2 The emerging market syndrome
2.7 Conclusion
3. STOCK MARKET THEORY
3.1 Introduction
3.2 The efficient market hypothesis and the present value model
3.2.1 The efficient market hypothesis and implications for stock market modelling
3.2.2 The present value model
3.3 Empirical implications of the present value model
3.3.1 The discount rate
3.3.2 Dividends and growth
3.4 Stock market asymmetry
3.5 Conclusion
4. A REVIEW ON EXISTING STOCK MARKET MODELS
4.1 Introduction
4.2 International studies
4.2.1 Studies evaluating stock market efficiency
4.2.2 Structural stock market models
4.3 South African studies
4.3.1 Studies on the efficiency of the South African stock market
4.3.2 Structural models of the South African stock market
4.4 Conclusion
5. A MARKOV SWITCHING REGIME MODEL OF THE SOUTH AFRICAN BUSINESS CYCLE
5.1 Introduction
5.2 The relationship between the business cycle and the yield spread
5.3 The econometric techniques
5.3.1 The Markov switching regime model
5.3.2 The logit model
5.4 Existing Markov switching regime business cycle models
5.4.1 Empirical Markov switching regime business cycle models with fixed transition probabilities
5.4.2 Empirical Markov switching regime business cycle models with time-varying transition probabilities
5.4.3 The yield spread as predictor of business cycles
5.5 Empirical analysis of the South African business cycle
5.5.1 Methodology
5.5.2 The estimated linear model
5.5.3 The estimated logit model
5.5.4 The estimated Markov switching regime model
5.6 Model selection
5.6.1 Comparing linear and Markov switching regime models
5.6.2 Comparing the estimated logit and Markov switching regime models
5.7 Conclusion
6. EMPIRICAL ESTIMATION OF THE SOUTH AFRICAN STOCK MARKET
6.1 Introduction
6.2 Data
6.3 Efficiency of the South African stock market
6.4 The cointegration equation
6.5 The short-run dynamics: an error correction model
6.6 Policy implications
6.7 Conclusion
7. COMPARING MODELS AND FORECASTS OF THE LEVEL AND TURNING POINTS OF THE SOUTH AFRICAN STOCK MARKET
7.1 Introduction
7.2 Modelling the level of the stock market
7.3 Modelling turning points in the stock market
7.4 Conclusion
8. SUMMARY AND CONCLUSION
8.1 Introduction
8.2 Modelling approach
8.3 Contributions of this study
8.4 Results
8.5 Conclusion
REFERENCES 
APPENDICES
GET THE COMPLETE PROJECT