AN ECONOMIC ANALYSIS OF FACTORS AFFECTING SOUTH AFRICAN EQUITY RETURNS By JUSTIN BEUKES Submitted in partial fulfilment of the requirements for the degree of Baccalaureus Commercii Honores (Economics) IN THE FACULTY OF BUSINESS AND ECONOMIC SCIENCES AT THE NELSON MANDELA METROPOLITAN UNIVERSITY Supervisor Mrs. D. Du Preez JANUARY 2009
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
AN ECONOMIC ANALYSIS
OF FACTORS AFFECTING SOUTH AFRICAN EQUITY RETURNS
By
JUSTIN BEUKES
Submitted in partial fulfilment of the requirements for the degree of
Baccalaureus Commercii Honores (Economics)
IN THE FACULTY OF BUSINESS AND ECONOMIC SCIENCES
AT THE NELSON MANDELA METROPOLITAN UNIVERSITY
Supervisor Mrs. D. Du Preez
JANUARY 2009
I
ACKNOWLEDGEMENTS
I would like to express profound gratitude to the following individuals for their involvement in this dissertation:
Deborah Du Preez, my supervisor, for her continued support throughout the year. Thank you for helping me through a difficult year. Without you, this would not have been possible.
Mario Du Preez, for being willing to always give a hand. Your guidance on the proposed topic was invaluable. Your enthusiasm in economics also rubbed off on me.
Leann Cloete, for being all that a friend can ask for. You did not once doubt my abilities.
Marius Wolmarans, for having a big impact on my life. Thank you for steering me in the right direction.
I am as ever, especially indebted to my family. Thank you for the love, support and counsel I could not do without.
II
EXECUTIVE SUMMARY
The efficient market hypothesis declares that the efforts of investors who attempt to
gain returns on share markets is futile. The current price of a share is said to reflect all
available information (Fama 1970) and that analysing available information with the
goal of earning a return will be unsuccessful.
Nonetheless, share price valuation has sought much attention because of the potential
gains to be realised by investors. The search for ways of reading the share market has
brought about two different approaches to share price valuation, namely, technical and
fundamental analysis. Technical analysis is a short term approach to valuation and is
the study of the action of the market (Edwards & Magee 2001: 4). That is, only the
prices of the companies concerned are studied ignoring forces outside the market.
Fundamental analysis is the second approach to share price valuation; it is basically
the study of value (Bodie et al., 2008: 569). A company’s value is determined by
examining virtually every aspect possible. The macroeconomy is one aspect that is
evaluated in order to determine the company’s macroeconomic environment.
This relationship, however, between certain macroeconomic variables and share
prices is not well established. There is no generally accepted asset pricing model to
explain this link (Asprem 1989). This is particularly the case for South Africa’s
Johannesburg Securities Exchange (JSE), as research on the JSE is limited.
The dissertation takes on a fundamental approach instead of a technical one to share
price valuation, to see whether fundamentals drive the JSE. It examines the
relationship between four macroeconomic variables and the JSE. The variables
considered are real activity, proxied by real GDP growth, the exchange rate as proxied
by the real effective exchange rate, inflation expectations as proxied by inflation
expectations for one year ahead, and the interest rate as proxied by the prime overdraft
interest rate.
An empirical analysis was carried out in order to determine if any of these four
variables help explain movements in the JSE All Share Index (ALSI).
It was hypothesised that an increase in real activity results in higher earnings and
therefore an increase in the ALSI, a depreciation of the currency improves firm’s
III
competitive position, thus increasing earnings and consequently increasing the ALSI,
higher inflation expectations are associated with increased interest rates which results
in a decrease in the ALSI, and lastly higher interest rates would lead to higher returns
through higher than usual short term inflows, thus improving the performance of the
ALSI.
The resultant model had some problems, i.e. it suffered from serial correlation. It
could not , however, be corrected for, as serial correlation still existed after running
the generalized least squares (GLS) equation using the AR(1) method.
Another attempt to correct for serial correlation was to re-specify the equation,
lagging the ALSI by one period and substituting inflation expectations for one year
ahead with current inflation expectations. This did not, however, solve the problem of
serial correlation and made all the variables insignificant. The initial regression was
thus reported on.
It was found that inflation expectations and the interest rate were statistically
significant variables in helping to explain movements in the ALSI. The other two,
namely GDP growth and the exchange rate were found to be insignificant.
Further research should thus be undertaken in order to identify possible significant
variables that will help to give investors a better understanding of the movements in
the ALSI, using the fundamental approach to share price valuation.
IV
LIST OF TABLES
Page
Table 2.1: Comparison of industry profitability ratios in South Africa
(2004 - 2005) 16
Table 3.1: Variables used in estimation 27
Table3.2: Regression output of equation 3.2 30
Table 3.3: Unadjusted R-squares and VIFs of independent variables 32
Table 3.4: The first-order serial correlation coefficient 34
Table 3.5: Generalized least squares AR (1) method 35
Table 3.6: White heteroskedasticity results 36
V
LIST OF FIGURES
Page
Figure 2.1:Bar chart and Point-and-figure chart 12
Figure 2.2:How resistance forms and How support forms 12
Figure 2.3:Dow Theory trends 14
Figure 2.4:Moving average for Microsoft as of 18 January 2005 15
Figure 2.5:Industry cyclicality 17
TABLE OF CONTENTS
Page
ACKNOWLEDGEMENTS I
EXECUTIVE SUMMARY II
LIST OF TABLES IV
LIST OF FIGURES V
CHAPTER ONE: INTRODUCTION
1.1 INTRODUCTION AND PRELIMINARY LITERATURE REVIEW 1
1.2 MODELS USED FOR PREDICTION PURPOSES 1
1.2.1 Technical Analysis 1
1.2.2 Fundamental Analysis 2
1.2.3 Fundamental Analysis Research 2
1.3 THE MACRO-ENVIRONMENT 3
1.3.1 Real Activity 3
1.3.2 Exchange Rates 4
1.3.3 Inflation Expectations 4
1.3.4 Interest Rate 5
1.4 PROBLEM STATEMENT 5
1.5 OBJECTIVES OF THE STUDY 6
1.6 RESEARCH HYPOTHESES 6
1.7 RESEARCH METHODOLOGY 6
1.8 ORGANISATION OF THE DISSERTATION 7
CHAPTER TWO: ALTERNATIVE APPROACHES TO SHARE
PRICE VALUATION
1.0 INTRODUCTION 8
1.1 THE THEORY UNDERLYING THE EFFICIENT MARKET
HYPOTHESIS 8
1.1.1 The Fair Game Model 9
1.1.2 The Submartingale Model 9
1.1.3 The Random Walk Model 9
1.1.4 The Forms of EMH 10
1.2 ALTERNATIVE APPROACHES TO SHARE PRICE VALUATION 10
1.2.1 Technical Analysis 11
1.2.2 Fundamental Analysis 15
2.4 A FUNDAMENTAL ANALYSIS OF SHARE PRICE VALUATION IN SOUTH AFRICA 20
2.4.1 Real Activity 21
2.4.2 Exchange Rate 22
2.4.3 Inflation Expectations 22
2.4.4 Interest Rate 23
2.5 CONCLUSION 24
CHAPTER THREE: AN EMPIRICAL ANALYSIS OF SHARE PRICE
VALUATION USING A FUNDAMENTAL
ANALYSIS APPROACH
3.1 INTRODUCTION 25
3.2 DATA SOURCES USED 26
3.2.1 Data Description 27
3.2.2 The Econometric Method of Estimation 28
3.3 AN EMPIRICAL APPROACH TO SHARE VALUATION
USING FUNDAMENTAL ANALYSIS 29
3.3.1 Model Structure 29
3.3.2 Presentation and Analysis 30
3.3.3 The Overall Fit of the Estimated Model 31
3.3.4 Violations of the Classical Linear Regression Model 31
3.4 CONCLUSION 37
CHAPTER FOUR: CONCLUSION
4.1 INTRODUCTION 39
4.2 GENERAL FINDINGS 39
4.3 CONCLUDING REMARKS 40
LIST OF SOURCES 42
CHAPTER ONE: INTRODUCTION
1.1 INTRODUCTION AND PRELIMINARY LITERATURE REVIEW
Conventional wisdom in economics suggests that share price changes are highly
unpredictable. This unpredictability is mainly caused by the fact that the fundamental
operation of the share market is very well understood. In essence, share prices exhibit
a random walk – this means that if shares did well last week, they are no more likely
to either do well or do poorly this week than at any other time. This so called random
walk is perceived by economists to be an indication of market efficiency and is
termed the efficient market hypothesis (EMH).
Market efficiency is theorised in three forms, namely, the weak, the semi-strong and
the strong. The weak form claims that share prices reflect all information from the
past, and thus an investor cannot predict future share prices on the basis of past share
prices. The semi-strong form argues that all publicly available information is already
represented in the share price; this information includes the quality of management,
fundamental data on the product, and so forth. This information cannot be used for
predictive purposes either. Finally, the strong form states that share prices reflect all
available information i.e. even unpublished information, and this (insider) information
cannot be used for predictive purposes (Bodie, Kane & Marcus 2007: 246) .
Notwithstanding the unpredictable nature of share prices, economists and those
intimately involved with share markets, still require processes or procedures to
explain and predict share market behaviour.
1.2 MODELS USED FOR PREDICTION PURPOSES
In the search for predictive models of share prices, two alternative models have come
to the fore, namely, fundamental analysis and technical analysis.
1.2.1 Technical Analysis
Technical analysis is basically the making and interpretation of share charts where the
analysis of averages and moving averages of share prices takes place. Technicians
study records of past share prices in order to find patterns they can profit from in the
1
future. They believe the market is 10 percent logical and 90 percent psychological.
They subscribe to the castle-in-the-air theory. This theory suggests that the crowd of
investors is likely to build their hopes during times of optimism into castles in the air.
The technician attempts to anticipate the crowd behaviour and thus buy before the
crowd makes their move (Malkiel 2003: 127).
1.2.2 Fundamental Analysis
In fundamental analysis, investors believe the market is 90 percent logical and 10
percent psychological. This model is based on the firm-foundation theory; the value
of a share is based on the present value of all future dividends. This value is known as
the intrinsic value. If the intrinsic value is greater than the quoted price the investor
should buy (Malkiel 2003:127). Fundamental analysts start their analysis with a study
of past earnings. In addition, an analysis of the management, the firms’ position in the
industry, and the prospects of the industry are taken into account (Bodie et al.,
2007:247). Fundamental analysts attempt to take all factors into consideration that
affect share prices, both firm specific and those pertaining to the macro-economy.
1.2.3 Fundamental Analysis Research
Asset prices are often thought to react to fluctuations in macroeconomic variables.
Returns on shares have a complicated relationship to macroeconomic variables.
Accordingly, research has being undertaken to understand this complexity.
Previous research by Solnik (1984: 69), was undertaken to find a correlation between
monetary variables and share prices. Variables considered were, interest rates,
inflation and the exchange rate. According to Solnik, the specific influence of
international monetary variables such as exchange rates is weak in comparison to
domestic variables such as changes in inflation expectations and interest rates. This
suggests that domestic variables are more influential.
Asprem (1989: 590) investigates the relationships, in ten European countries, between
the countries’ respective major share index and various macroeconomics variables.
This research showed that changes in share prices are positively correlated to certain
measures of real economic activity. These measures included a) industrial production,
b) real gross national product, c) gross capital formation, and d) exports. Positive
correlations were shown to exist between changes in the share indices and the United
2
States yield curve, as well as the M1 monetary aggregate and lagged values of share
indices themselves. Negative correlations are found between changes in share indices
and employment (a measure of real activity), the exchange rate, imports, inflation and
interest rates.
1.3 THE MACRO-ENVIRONMENT
Based on work done by Asprem (1989), the macroeconomic variables considered in
this dissertation include: a) real activity, b) exchange rates, c) interest rates and d)
inflation expectations.
1.3.1 Real Activity
In fundamental analysis, the constant growth model, also known as the Gordon
growth model, is used in calculating the intrinsic value of a firm. Value is based on
the future series of dividends that grow at a constant rate. Given a dividend per share
that is payable in one year, the required rate of return , and the assumption that the
dividend grows at a constant rate in perpetuity, the model solves for the present value
of the infinite series of future dividends (Howells & Bain 2005: 347).
Asprem (1989: 593), who assumes rational markets, suggests that asset prices should
reflect expectations of these future earnings, which are inclined to be influenced by
measures for real activity.
The perceptions of investor’s on future dividends must be responsive to changes in
the outlook for the economy as a whole. The movement from boom to recession, for
example, will lead to a demoted perception of dividend forecasts. Thus in the Gordon
growth model, this will lead to a reduced dividend in the numerator, and therefore a
downgraded present value.
The business cycle represents different levels of real activity. The different stages of
the cycle will exhibit differing levels of performance. In an expansionary phase,
performance will be greater, and expectations of future earnings thus increase. The
opposite is true for a recession, where expectations will be lower (Bodie et al., 2005:
577).
3
These expansionary phases can be accounted to market growth, thus the gross
domestic product (GDP) growth rate will be used as a measure of this growth or
increase in real activity.
1.3.2 Exchange Rates
According to Phylaktis and Ravazzolo (1998), the adoption of more flexible exchange
rate regimes by developing countries has increased the volatility of foreign exchange
markets and the risk associated with such investments. The choice of currency
denomination is an important dimension in the overall portfolio decision of the
investor. Exchange rates volatility also affects real activity in that it changes the prices
of imports and exports.
A depreciation of a currency improves the competitive position of domestic
industries. The earnings of an export or international firm will increase when the
currency depreciates, as domestic goods will be cheaper relative to other international
goods (Asprem, 1989: 596). However, domestic firms that need to import capital
equipment will experience a negative impact on their share prices.
Any investor should be concerned about how their domestic capital market reacts to
international monetary disturbances such as exchange rate fluctuation. Solnik (1984:
71) maintains that the international investor who uses his home currency to value his
portfolio measures return as the sum of his assets’ return, in local currency, plus any
currency movements. The portfolio thus bears both market and currency risk. Leaving
market risk aside by way of diversification, the investor still needs to pay attention to
reactions of share prices to fluctuations in the currency of measure.
It is thus expected that a negative relationship exists between exchange rates and
share prices, due to the impact on domestic industries.
1.3.3 Inflation Expectations
The higher inflationary expectations are, the more likely one’s real returns on share
investments are going to be negatively affected.
Higher inflationary expectations also bring about higher interest rate expectations.
Central banks are generally forward looking in the application of monetary policy,
implying that if they expect higher inflation in the future, then they will apply
4
restrictive monetary policy today. This usually involves an increase in the interest
rates. Indirectly, inflationary expectations in most cases, leads to increases in interest
rates, which generally suppresses share market performance.
1.3.4 Interest Rate
The most considerable source of market wide influences can be contributed to interest
rates or expectations of interest rates. The required rate of return in the denominator of
the Gordon growth model is the sum of a risk-free rate and a risk premium derived
from the market’s current pricing of risk in general and the firm’s relative risk
characteristics. An official change in interest rates causes a change in the risk free rate
which will cause the required rate of return to change. An increase in the denominator
caused by increased interest rates will reduce the present value of the shares in
question; the opposite is true for a decrease (Howells and Bain, 2005: 356).
However, investors consider how much interest can be earned in all investment
opportunities (Malkiel 2003: 112). The risk free rate provides a base for all risky
assets. Once the interest rate changes the opportunity cost for investors in equity
markets changes (Asprem 1989: 598).
Interest rates are thus expected to be positively related to share prices according to the
opportunity costs investors undertake.
1.4 PROBLEM STATEMENT
Investors are observers of numerous factors that might affect return on equity. Returns
on shares have a complex relationship with macroeconomic variables. Thus there is no
consensus on a generally accepted asset pricing model that explicitly takes economic
variables into account (Asprem, 1989: 589).
This complex linkage of macroeconomic variables when related to shares on the
Johannesburg Securities Exchange (JSE) has not found much attention. Research on
this topic with regards to the South African equity market is thus limited. A model
needs to be developed to explain macroeconomic variations in share prices on the JSE.
5
1.5 OBJECTIVES OF THE STUDY
(a) Primary objective:
The primary objective of this work will be to research the link between certain
macroeconomic variables and share prices on the JSE.
(b) Secondary objective:
The secondary objective is to provide a simple and useful model that can be used by
investors and prospective investors in the South African equity market, to predict the
general movement of future equity returns.
1.6 RESEARCH HYPOTHESES
The following hypotheses will be tested:
Increased real activity in the form of GDP growth results in higher earnings and
therefore an increase in the JSE All Share Index (ALSI).
An expected depreciation in the exchange rate places domestic firms in a greater
competitive position, owing to cheaper exports, thus increasing earnings and
therefore increasing the ALSI.
Higher inflation expectations will have a negative impact on the ALSI.
Investors see a chance of higher returns with higher interest rates thus boosting the
short term inflows and increasing the ALSI.
1.7 RESESEARCH METHODOLOGY
Data will be obtained through secondary sources using the following methods:
(a) Historical method:
This involves obtaining data from published sources such as journals, research
reports, articles and the internet.
(b) Analytical method:
This involves the drawing up of an econometric model to establish statistical
relationships between the ALSI and certain macroeconomic variables investigated.
6
1.8 ORGANISATION OF THE DISSERTATION
Chapter two presents a literature overview, which explains both technical and
fundamental analysis. Fundamental analysis is further expanded on with regards to the
four factors considered, namely, real activity, exchange rates, inflation expectations
and interest rates. Chapter three entails an econometric analysis of the correlation
between the considered factors and the ALSI. Chapter four involves an overall
conclusion and highlights key points from each section.
7
CHAPTER TWO: ALTERNATIVE APPROACHES TO SHARE PRICE
VALUATION
2.1 INTRODUCTION
The potential gains which the share market holds to those who read it correctly are
colossal; then again, the potential losses to those who don’t can be severe. It has
drawn people from all walks of life in an endeavour to reap these enormous gains
(Edwards & Magee, 2001: 3). The share market has been extensively studied in
finding a way to predict share price behaviour. Early research on share prices could
find no predictable patterns (Kendall 1953). Prices seemed to behave randomly; there
was no economic rationale to explain this until the development of the efficient
market hypothesis (EMH). The EMH explained the random movement in prices as the
result of a well operating market (Bodie et al., 2008: 357). Despite this theory of
market efficiency, two divergent approaches to predicting share prices have evolved,
namely, technical analysis and fundamental analysis.
The purpose of this chapter is to expand on the EMH and the two approaches
mentioned above, and give a more comprehensive explanation of each. The grounding
theory of the EMH is illustrated by Fama (1970) in three models namely, a) the Fair
Game model, b) the Submartingale model and c) the Random Walk model. The
differing versions of the EMH are explained by Bodie et al. (2008). Technical
analysis will be divided into two areas, according to Teweles and Bradley (1982:
373), namely; a) patterns on price charts, and b) trend-following methods.
Fundamental analysis is explained by Bodie et al. (2008) using macroeconomic and
equity analysis. Lastly, macroeconomic factors based on work done by Asprem
(1989), namely: a) real activity, b) exchange rates, c) interest rates and d) inflation
expectations, are applied to share price valuation in South Africa.
2.2 THE THEORY UNDERLYING THE EFFICIENT MARKET HYPOTHESIS
The EMH plays an important part in financial economic literature. An asset market is
said to be efficient if the asset price reflects all information. If this is true, investors
are wasting their time in an attempt to earn abnormal returns.
8
Fama (1970) suggested the following models for testing stock market efficiency: the
Expected Return or Fair Game model, the Submartingale model, and the Random
Walk model.
2.2.1 The Fair Game Model
The fair game model states that investors cannot achieve above average returns based
on historic information because such information is fully integrated into the share
price (Bhatti et al., 2006: 230). Fama (1970: 384) defines more exactly what is meant
by the term “fully reflected”. He hypothesizes that equilibrium prices are generated in
a two parameter world. These prices are conditional on some information set, and the
equilibrium expected return on a share is a function of its risk. This information is
fully utilised in determining equilibrium expected returns i.e. “fully reflected”.
Furthermore, the model hypothesises, that expected profits or returns in excess of
equilibrium are a fair game with respect to this information set. What is meant by
“fair game” is that expected profits or returns are zero.
2.2.2 The Submartingale Model
This model is similar to the fair game model, but the dissimilarity is that the expected
return is positive. Prices are expected to increase over time. Returns on investments
are expected to be positive due to the risk involved in capital markets. The expected
value of next period’s price, as forecast on the basis of the information set, is equal to
or greater than the current price. If the expected value of next period’s price is equal
to the current price, then the price sequence follows a martingale (i.e. zero expected
return) (Fama 1970:386).
2.2.3 The Random Walk Model
Two hypotheses constitute the random walk model. Firstly, the statement that the
current price of a share “fully reflects” available information is assumed to imply that
successive price changes are independent. Secondly, it is assumed that successive
price changes are identically distributed (Fama 1970: 387). The random walk model
states more than the fair game model. The fair game model states that the mean return
of the next period is independent of the current information set, however, the random
9
walk model, in addition, states that the entire distribution is independent of the current
information set.
2.2.4 The Forms of EMH
The EMH can now be described in more detail in accordance with the available
information that is reflected in the price. Fama (1970: 383) classified the information
set into three groups and put three forms of EMH forward, depending on the
definition of the relevant information set. These three forms are the weak, semi-strong
and strong forms.
The weak form hypothesis is the lowest form of efficiency. It maintains that share
prices already reflect all information that can be obtained by examining market
trading data, for example, past prices and trading volume. This form implies that past
data cannot be used for predictive purposes (Bodie et al., 2008: 361). The information
set applicable in the semi-strong form is publicly available information. This
information includes, in addition to past prices; annual reports, quality of
management, interest rates, information on money supply and the exchange rate, to
name a few. Consequently, investors are unable to reap superior returns from
analysing this data available to the public (Bodie et al., 2008: 361). The strong form is
concerned with whether given investors or groups have monopolistic access to any
information relevant to price formation (Fama 1970: 383). This version of the
hypothesis is extreme in that both private (inside information) and public information
is reflected in the price (Bodie et al., 2008: 361).
2.3 ALTERNATIVE APPROACHES TO SHARE PRICE VALUATION
The two different approaches to share price valuation include technical and
fundamental analysis. These theories have been put forward despite the EMH, which
states that both past and public information cannot be used for predicting future share
prices. Technical analysis is concerned with the market and the reading of charts (i.e.
past information), whereas fundamental analysis takes a scope of variables that lie
outside the market into account (i.e. public information).
10
2.3.1 Technical Analysis
Technical analysis is the study of the action of the market and ignores the study of the
goods in which the market deals (Edwards & Magee 2001: 4). In contrast,
fundamental analysis looks carefully at fundamental economic and political
conditions – forces inside and outside of the market. Technicians do not reject the
value of fundamental information, but believe prices only gradually close in on
intrinsic value (Bodie et al., 2008: 407). Technical analysts, primarily, are short run
traders and only seek capital gains. The technical approaches of patterns on price
charts and trend following methods, will be discussed below (Teweles & Bradley
1982: 372).
Patterns on Price Charts
Among the most oldest and popular approaches is the use of patterns on price charts.
Chartists believe that patterns repeat themselves and that charts can be used to
forecast significant price movements (Teweles & Bradley 1982: 373). Charts are used
to record historical movements of share prices and to forecast future movements
(Badger & Coffman 1967: 187). The two most commonly used charts are bar charts,
and point and figure charts.
An illustrative comparison of a bar chart and a point and figure chart is given in figure
2.1 for the same price fluctuation and time period. Bar charts indicate time on the
horizontal axis and price on the vertical axis in figure 2.1 (a). Price is represented by a
vertical line for a specified length of time, usually a day. The bar drawn indicates the
price range from high to low. Tick marks are added to indicate the opening and the
close of the market, and the midrange may be specified. Each period is plotted
chronologically from left to right. Additional data may be added that may be of
importance to the trader, such as volume (Teweles & Bradley 1982: 374).
Point and figure charts in contrast, in figure 2.1 (b), take no consideration of time or
volume – they are only designed to indicate price change. Significant changes,
considered by the individual trader, are represented by each box. Increases of this
amount are indicated by an X. Each successive rise is indicated by an X on top of the
previous X. If there is a decrease of the selected amount, an X is entered in the
following column and one box down. This is called a reversal. To make interpretation
11
easier, rises may be indicated by X’s and falls by O’s. Closing prices can also be
blacked in (Teweles & Bradley 1982: 375).
Figure 2.1: (a) Bar chart (b) Point-and-figure chart ( from Teweles & Bradley 1982 ).
Since technicians are only interested in the action of the market, much analysis is
centered around the interaction between demand and supply for a share. Increasing
prices are indicative of more demand than supply, and decreasing prices indicate the
opposite. Accordingly, a support level is a price at which a considerable increase in
the demand for a share is to be expected, whereas a resistance level is the price where
a substantial increase in the supply of a share develops. At these levels, a relatively
large amount of shares change hands (Badger & Coffman 1967: 198).
Figure 2.2: (a) How resistance forms (b) How support forms (from Badger &
Coffman 1967).
In panel (a) of figure 2.2, for example, a number of investors may have purchased
shares between R20 and R22. Say the share slips to R18 then recovers to the R20-R22
range. Holders hope the share will rise above R22, but some are happy to get out even
and start to sell when the share hits R22. This liquidation prevents the share from
going above R22, thus R22 forms as a resistance line. Perhaps this pressure from
12
Pri
ce
Pri
ce
Weeks
(a)
Weeks
(b)
Pri
ce
Pri
ce
(a) (b)
liquidation overwhelms new buying of the share, and the share breaks through the
lower support line. Then this range of R20-R22 (a congestion range) is interpreted as
a resistance level. In panel (b) of figure 2.2, the opposite is illustrated. Buyers in the
range R20-R24 are content with their purchase. When the share drops to R20, buyers
are attracted, and the resistance level of R24 meets sustained pressure. Eventually, the
prices break out on the upside, the area R20-R24 becomes known as a support level
(Badger & Coffman 1967).
Although many statistics books show how to construct charts, none present any
statistics indicating that charts will probably lead to significant profits over time
(Teweles & Bradley 1982: 375).
Trend Following Methods
A great deal of technical analysis is the identification of trends in market prices.
Technical analysts believe that once a trend is established, it is more likely to continue
than to reverse (Teweles & Bradley 1982: 377). This is basically the search for
momentum in market prices. Momentum can be absolute - where a trader searches for
upward or downward price trends, or it can be relative - when an analyst seeks to
invest in one sector over another (Bodie et al., 2008: 407).
a) Dow Theory
Trend analysis has its origins in the Dow Theory, established by Charles Dow. The
Dow theory hypothesises three forces working at the same time on share prices,
namely the primary, secondary and tertiary trend. The primary trend is the long term
change of prices over a number of months to a number of years. The secondary (or
intermediate) trends are brought about by short term deviations of prices from the
underlying trend line. The tertiary (or minor trends) are fluctuations on a daily basis,
which are insignificant (Bodie et al., 2008: 408).
13
.
Figure 2.3: Dow Theory trends (from Bodie et al., 2008).
Figure 2.3 above points out the characteristic components of the Dow Theory. The
primary trend here is upward, the intermediate trend has temporary deviations for a
few weeks and the minor trend has no long run impact (Bodie et al., 2008: 408).
b) Moving averages
The most usual trend following device is the moving average. A moving average adds
a new term periodically, for instance daily or weekly, and at the same time drops the
oldest term, thus recalculating the average per day or week (Teweles & Bradley 1982:
377).
The moving average takes in older and higher prices, thus it will be above current
prices after a period that has been experiencing falling prices. When prices have been
rising, the moving average will be below the current price. A break of the market
price (the irregular curve) through the moving average line (the smoothed curve) from
below, as at point A in figure 2.4, is a bullish signal, as it implies a shift from a falling
trend to a rising trend. The opposite is true for point B, the market line cuts the
moving average line from above and is thus a bearish signal.
The followers of trends must decide on the average lengths of time to utilize and what
events induce market action. The method can be checked with respect to past markets
without risking capital as the device allows them to get out of the market as well as
into it. Traders will profit from a major fluctuation in the market, and not lose much
money before the position is abandoned (Teweles & Bradley 1982: 379).
14
Pri
ce
Time
Figure 2.4: Moving average for Microsoft as of 18 January 2005 (from Bodie et al.,
2008).
2.3.2 Fundamental Analysis
The alternative method to share price valuation is fundamental analysis. Fundamental
analysis not only looks at the action of the market but also examines variables outside
of the market. The core of fundamental analysis is the analysis of the determinants of
value. Value is analysed at a micro and macro level. Analysis takes place at firm
level looking at the financials and even the personal characteristics of management.
However, the individual firm is tied to the broader economy, and thus
macroeconomics factors can have an influence on its future prospects as well (Bodie
et al., 2008: 569).
Macroeconomic and Industry Analysis
If the fundamental analyst first reviews the macro-economy and the specific industry,
a top down approach is being taken. This approach is a way of evaluating a firm’s
prospects by looking at the bigger picture first. After the macroeconomic influences
are considered, the fundamental analyst examines the firms’ position in the particular
industry (Bodie et al., 2008: 571).
Macroeconomic factors can be organised into demand and supply shocks. Demand
shocks are events that affect the demand for goods and services in an economy,
whereas supply shocks are events that influence production capacity and costs.
Demand shocks usually occur by aggregate output moving in the same direction as
interest rates and inflation, while supply shocks are the converse, with output moving
in the opposite direction of inflation and interest rates (Bodie et al., 2008: 574).
15
Pri
ce (
in m
illi
ons)
Time
Industry analysis, similarly with macroeconomic analysis, is important in evaluating a
firms’ performance, as it is generally impossible for a firm to do well in its industry
when the industry in question is suffering. This importance of selecting the correct
industry is illustrated in table 2.1. The profitability ratio1 for industries in South Africa
for 2004 and 2005 are given. In 2005, the profitability ratio for trade was only 4%
whereas real estate and other business services (excluding financial intermediation
and insurance) was 15%. An investor must choose the correct industry, as profitability
between differing industries can be large.
Table 2.1: Comparison of industry profitability ratios in South Africa (2004 –
2005)
Industry Profitability ratio
2004 2005
Forestry and fishing - 0.10
Mining and quarrying 0.06 0.07
Manufacturing 0.06 0.08
Electricity, gas and water supply 0.10 0.10
Construction 0.03 0.05
Trade 0.03 0.04
Transport, storage and communication 0.05 0.11
Real estate and other business services, excluding financial intermediation and insurance
0.15 0.15
Community, social and personal services, excluding government institutions
0.08 0.10
All industries 0.06 0.07
Source: Statistics South Africa (2005)
After the state of the macroeconomy is forecast, the analyst must determine the
implications of the forecast for specific industries. Industries have differing
sensitivities to the business cycle. Three factors determine the sensitivity of a firm’s
earnings to the business cycle. Firstly, is the sensitivity of sales. Necessities show
little sensitivity, as demand for these goods remain intact during recessions. Also,
industries for which income is not a crucial determinant of demand, for example, have
low sensitivity. Secondly, operating leverage determines sensitivity. This refers to the
ratio of variable to fixed costs. A firm with more variable costs with respect to fixed
costs is less sensitive to the conditions of business. During a recession, costs can be
reduced as output falls. Whereas, a firm with high fixed costs is more sensitive as
1 Net profit after providing for company tax divided by turnover
16
costs cannot be reduced. Operating leverage is measured by how sensitive profits are
to changes in sales. A degree of operating leverage of greater than 1 represents some
operating leverage. The third factor that influences sensitivity is financial leverage.
The use of debt incurs interest payments which must take place regardless of business
conditions. These payments represent a fixed cost similar to the operating leverage
case (Bodie et al., 2008: 586).
Differing sensitivity to the business cycle of the passenger car and cigarette industries
is demonstrated in figure 2.6 below. Cigarettes (the black curve in figure 2.6) will be
consumed regardless of whether economic times are good or bad and thus are less
sensitive. This is illustrated by the fairly smooth movement within the industry.
Durable goods, however, such as passenger cars (grey curve in figure 2.6) are
sensitive to the swings of the business cycle (Bodie 2008: 586). This is illustrated
through the large fluctuations in the passenger car industry.
Figure 2.5: Industry cyclicality (from Bodie et al., 2008)
Equity valuation
This valuation is at the firm level. Here, individual firm’s shares are valued according
to dividend models and are compared using price-earnings ratios.
17
a) Dividend Models
Fundamental analysts are always on the search for mispriced shares using information
concerning the current and prospective profitability of a company to calculate its fair
value (Bodie et al., 2008: 603). Dividend models are the most accepted and
conventional approaches used by fundamentalists for asset valuation. Investors in
shares expect a return consisting of dividends and capital gains or losses (Bodie et al.,
2008: 605).
This approach states that the present value of an asset is the sum of its future earnings,
each discounted at an appropriate rate that takes time and risk into account (Howells
& Bain 2005: 346). The price of a share at time period zero P0 is the present value
PV of the expected future dividend streamD1. The expected capital gain realised upon
the sale of the share is included in the expected future dividend stream since its size
also depends on the present value of this stream. The approach assumes that expected
dividends grow at a constant rateg, thus eliminating the problem of forecasting an
infinite number of dividends. Finally, the required rate of return is given byk . Taking
these elements into account gives the constant growth model, also known as the
Gordon growth model, in equation 2.1 (Howells & Bain 2005: 347):
INTRATE 2.88E+08 2.06E+08 1.398091 0.1801INTRATE^2 -3385866. 17255032 -0.196225 0.8468R-squared 0.583059 Mean dependent var 38391090Adjusted R-squared 0.239697 S.D. dependent var 70410037S.E. of regression 61394299 Akaike info criterion 39.00851Sum squared resid 6.41E+16 Schwarz criterion 39.69557Log likelihood -609.1362 F-statistic 1.698085Durbin-Watson stat 1.203038 Prob(F-statistic) 0.149121
The hypotheses are stated as follows:
36
H 0 : No Heteroskedasticity
H A : Heteroskedasticity is present
The null hypothesis is only rejected if the test statistic is greater than the critical value
obtained from the Chi-square distribution tables.
The results of the white test are displayed in table 3.6 above. The Obs*R-squared
value of 18.66 is the White’s test statistic. It is computed as the number of
observations time the R2 from the test regression. The test statistic has a chi-square
distribution with degrees of freedom equal to the number of slope coefficients in table
3.6. Thus, with 14 degrees of freedom, the critical chi-square value is 23.7 at the 5 per
cent level of significance. Since the test statistic of 18.66 is less than the critical value
of 23.7, the null hypothesis cannot be rejected, and it is thus concluded that no
heteroskedasticity is present.
3.4 CONCLUSION
Theory suggests that the four variables considered in this dissertation should relate to
the ALSI as follows. Higher inflation expectations, according to the Gordon growth
model, should lead to a higher discount rate and thus current share prices
should drop. Theory on the exchange rate was inconclusive (remembering though that
not much research has been done in this area). The portfolio balance model suggests
that a currency depreciation will have a negative effect on share prices. However,
looking at the effect of a currency depreciation on domestic industries, share prices
could rise. Real activity, represented by GDP growth, should have a positive
relationship with the ALSI. Increased GDP should be reflected by higher earnings and
thus higher share prices. Lastly, using the Gordon growth model again, higher
interest rates increase the discount rate and thus share prices should decrease.
Short term inflows, however, from perceived higher returns, associated with higher
interest rates can increase share prices.
In the initial regression (table 3.2) all coefficients obeyed these relationships, except
for the coefficient of the interest rate variable. The overall fit of this regression was
37
not that impressive, with only 27 per cent explanatory power, according to the
adjusted R-squared.
The model was tested for multicollinearity, serial correlation and heteroskedasticity to
see if it conformed to the assumptions of the classical model. With respect to
multicollinearity, only inflation expectations showed severe multicollinearity, with a
VIF of 5.4. Since inflation expectations were statistically significant, no remedial
action was taken. Using the Durbin-Watson d test, positive serial correlation was
found. Therefore the GLS AR(1) method was run to solve the serial correlation.
However, the signs of the exchange rate variable and the interest rate variable
changed signs. Also the p-values for all the variables in table 3.5 show that the
coefficients are insignificant at the 5 per cent level. In an attempt to correct for this
serial correlation, the ALSI was lagged by one period and inflation expectations for
one year ahead was replaced by current inflation expectations. This new equation still
gave disappointing results when the GLS AR(1) method was run, with the outcome of
insignificant coefficients. Therefore the initial regression was kept. According to the
White Test, the initial model showed no heteroskedasticity.
38
CHAPTER 4: CONCLUSION
4.1 INTRODUCTION
The first chapter introduced the topic of research with a preliminary literature review.
The literature review gave an outline of the two models used for predictive purposes,
namely technical and fundamental analysis. Fundamental Analysis was expounded by
the macroeconomic factors that were considered, specifically real activity, exchange
rates, inflation expectations and the interest rate. The problem statement indicated that
a complicated relationship exists between share prices and macroeconomic factors,
and that research in this area for the Johannesburg Securities Exchange (JSE) was
limited. The objective of the study was to investigate the connection between the
performance of the All Share Index (ALSI) and the four macroeconomic variables and
to develop a model to explain these connections.
The second chapter was a theoretical examination of the two approaches to share
price valuation. Despite the efficient market hypothesis (EMH), the two models,
mentioned above, dominate the literature on share price valuation. Technical analysis
is a very vast subject, with many techniques been developed over the years of share
market research. In the dissertation this model was shortened to a discussion on
patterns on price charts and trend following methods. Fundamental analysis was then
explained, with a discussion on value, the macroeconomy and industry, and
equity. The fundamental model was used in the dissertation focusing only on the
four variables mentioned above.
Chapter two concluded with the question of whether shares reflect fundamental
economic factors. Subsequently, chapter three attempts to answer this question and to
find a better understanding of the correlations that exist between the macroeconomic
variables investigated and the ALSI.
4.2 GENERAL FINDINGS
The empirical analysis of chapter three was not as significant as expected. Asprem
(1989: 589) declared that no commonly accepted pricing model exists that only
39
takes macroeconomic variables into account. This declaration was investigated with
respect to the JSE, and it seems as though this is still the case.
A linear multivariate regression model was built using the ALSI as the dependent
variable and inflation expectations, the real effective exchange rate, GDP growth and
the prime overdraft rate as independent variables. The coefficients of inflation
expectations and the interest rate were the only variables that were statistically
significant at the 5 per cent level. According to the adjusted R-squared, all the
variables explained only 27 per cent of the variation in the ALSI. The model was
constructed to obey the classical linear regression model, thus it was checked for
multicollinearity, serial correlation and heteroskedasticity. A remedy was only sought
for serial correlation by estimating the generalized least squares (GLS) equation using
the AR (1) method. However, this did not adequately solve the problem of serial
correlation.
Another equation was regressed, lagging the ALSI for one period and substituting
inflation expectations for one year ahead with current inflation expectations. This was
done to see if better results could be found; in spite of this the results were no more
promising than the initial regression, and thus are not included here.
4.3 CONCLUDING REMARKS
The empirical analysis carried out in this dissertation encountered two major
difficulties. Firstly, not enough data could be found for inflation expectations,
therefore EViews included only 32 observations after adjusting the endpoints. It is
generally accepted that the more sample observations one has, the more reliable ones
estimates can be. Secondly, serial correlation could not be solved regardless of the
efforts made, as mentioned above. This dissertation thus concludes that only inflation
expectations for one year ahead and the interest rate are statistically significant in
explaining the variation in the ALSI. The presence of serial correlation, however,
caused the ordinary least squares (OLS) estimates of the standard errors to be biased,
leading to unreliable hypothesis testing (Studenmund 2006: 324). Given the
unreliable and biased results obtained, it is difficult to determine whether any of the
hypotheses stated in chapter one are true.
40
Increased real activity did lead to an increase in the ALSI, however this result was
insignificant in the initial model. A depreciation in the exchange rate did lead to better
performance on the ALSI, however, this result was also found to be insignificant (see
table 3.2). The interest rate showed a positive relationship with the ALSI, which
highlighted the importance of international investors in South Africa. Short term
inflows increased, leading to a better performing ALSI. This result was significant,
however, this was prior to correcting for serial correlation. Lastly, inflation
expectations had an expected negative relationship with the ALSI. This result was
also significant, but once again, did not account for possible serial correlation.
In conclusion, further research should be carried out in order to identify significant
variables that help explain the performance of the ALSI, under the fundamental
approach to share price valuation.
41
LIST OF SOURCES
Asprem, M. 1989. Stock Prices, Asset Portfolios and Macroeconomic Variables in Ten European Countries. Journal of Banking and Finance, Vol. 13, pp.589 – 612.
Badger, R.E. & Coffman, P.B. 1967. The Complete Guide to Investment Analysis. New York: McGraw-Hill.
Barr, G. & Kantor, B. (2002) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Barr, G. & Kantor, B. 2002. The South African Economy and its Asset Markets – An Integrated Approach. The South African Journal of Economics, Vol. 70(1), pp. 53 – 77.
Bethlehem, G. (1972) in Van Rensburg, P. 1995. Macroeconomic variables and the Johannesburg Stock Exchange: A multifactor approach. De Ratione, Vol. 9(2), pp.45-63.
Bhatti, M.I. Al-Shanfari, H. & Hossain, M.Z. 2006. Econometric Analysis of Model Selection and Model Testing. Ashgate Publishing, Ltd
Bodie, Z. Kane, A. & Marcus, A.J. 2005. Investments. New York: McGraw-Hill.
Bodie, Z. Kane, A. & Marcus, A.J. 2007. Essentials of Investments. 6th Ed. New York: McGraw-Hill.
Bodie, Z. Kane, A. & Marcus, A.J. 2008. Investments. 7th Ed. New York: McGraw-Hill.
Branson, W.H. & Henderson, D.W. (1985) in Fang, W. 2002. The effects of currency depreciation on stock returns: evidence from five East Asian economies. Applied Economics Letters. Vol 9, pp. 195-199.
Chen, N., Roll, R. & Ross, S. (1986) in Asprem, M. 1989. Stock Prices, Asset Portfolios and Macroeconomic Variables in Ten European Countries. Journal of Banking and Finance, Vol. 13, pp.589 – 612.
Correia, C.J. & Wormald, M.P. (1987) in Van Rensburg, P. 1995. Macroeconomic variables and the Johannesburg Stock Exchange: A multifactor approach. De Ratione, Vol. 9(2), pp.45-63.
Edwards, R.D. & Magee J. 2001. Technical Analysis of Stock Trends. 8th Ed. New York: Amacom
Fama, E.F. 1970. Efficient Capital Markets: A Review of Theory and Empirical Work. Journal of Finance, Vol. 25, pp. 383 – 417
Fama (1981). in Asprem, M. 1989. Stock Prices, Asset Portfolios and Macroeconomic Variables in Ten European Countries. Journal of Banking and Finance, Vol. 13, pp.589 – 612.
Fang, W. 2002. The effects of currency depreciation on stock returns: evidence from five East Asian economies. Applied Economics Letters. Vol 9, pp. 195-199.
Forssman, A. 2005. FTSE/JSE End of Day Indices Dissemination User Manuel. Online.
Available: http://ftse.jse.co.za/product_specs.jsp [Accessed: 25 November 2008]
Gultekin, N.B. (1983) in Van Rensburg, P. 1995. Macroeconomic variables and the Johannesburg Stock Exchange: A multifactor approach. De Ratione, Vol. 9(2), pp.45-63.
Howells, P. & Bain, K. 2005. The Economics of Money, Banking and Finance. 3rd Ed. Harlow: Prentice Hall.
Jefferis, K.R. & Okeahalam, C.C. (2000) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Jefferis, K.R. & Okeahalam, C.C. 2000. The Impact of Economic Fundamentals on Stock Markets in Southern Africa. Development Southern Africa, Vol. 17(1), pp. 23-51.
Jondeau, E. & Nicolai, J.P. (1993) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Keim and Stambaugh (1986) in Asprem, M. 1989. Stock Prices, Asset Portfolios and Macroeconomic Variables in Ten European Countries. Journal of Banking and Finance, Vol. 13, pp.589 – 612.
Kendall, M. (1953) in Bodie, Z. Kane, A. & Marcus, A.J. 2008. Investments. 7th Ed. New York: McGraw-Hill.
Levine, R. & Zervos, S. (1996) in Jefferis, K.R. & Okeahalam, C.C. 2000. The Impact of Economic Fundamentals on Stock Markets in Southern Africa. Development Southern Africa, Vol. 17(1), pp. 23-51.
Malkiel, B.G. 2003. A Random Walk Down Wall Street. New York: W.W Norton & Company.
Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Phylaktis, K. & Ravazzolo, F. 1998. Stock Prices and Exchange Rate Dynamics. Online. Available: http://www.cass.city.ac.uk/emg/workingpapers/stock_prices_and_exchange.pdf [Accessed: 21 March 2008]
Reilly, F.K. (1989) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Solnik, B. 1984. Stock Prices and monetary variables: The International Evidence. Financial Analysts Journal, Vol. 40, pp.69 – 73.
Studenmund, A.H. 2006. Using Econometrics: A practical guide. 5th edition. Cape Town:
Pearson Addison Wesley.
Teweles, R.J. & Bradley, E.S. 1982. The Stock Market. 4th Ed. New York: John Wiley and Sons.
Van Rensburg, P. (1995) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Van Rensburg, P. (1998) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Van Rensburg, P. (1999) in Moolman, E. & Du Toit, C. 2005. An Econometric Model of the South African Stock Market. Online. Available: https://www.up.ac.za/dspace/bitstream/2263/3031/1/Moolman_Econometric(2005).pdf [Accessed: 10 October 2008]
Van Rensburg, P. 1995. Macroeconomic variables and the Johannesburg Stock Exchange: A multifactor approach. De Ratione, Vol. 9(2), pp.45-63.