Investment guidelines based on future growth indicators

(1)

INVESTMENT GUIDELINES

BASED ON FUTURE GROWTH

INDICATORS

CHRISTO VORSTER

Dissertation submitted in partial fulfilment of the requirements for the degree

MASTER IN BUSINESS ADMINISTRATION at the NORTH-WEST UNIVERSITY

POTCHEFSTROOM CAMPUS

Study Leader: Prof Ines Nel

POTCHEFSTROOM

(2)

Investment guidelines based on future growth indicators

Abstract

The stock market is cited to be one of the greatest tools ever invented for building wealth. The relative small size of the ideal portfolio, consisting of 10 to 12 shares, reiterates the fact that share selection is absolutely crucial to portfolio success and ultimately the creation of personal financial independence.

The main objective of this study is to research, identify and develop investment guidelines based on possible future growth indicators of organisations listed on the JSE.

Various possible growth indicators are identified and are statistically related to typical growth metrics, namely sales turnover, sales turnover growth, average share price and average share price growth. These possible growth indicators, identified in Chapter 2 are:

• Operating profit margin. • Capital requirements ratio. • Return on equity (ROE). • Retention ratio.

• Financial leverage. • Net profit margin.

• Return on assets (ROA). • Total asset turnover. • Earnings per share (EPS). • Cash flow per share. • Inventory; and • Dividend yield.

This study finds cash flow per share, inventory and dividend yield to be the top 3 future growth indicators, based on extensive backward stepwise multiple regression modelling, covered in Chapter 3. The private investor can utilize these indicators to assist in identifying feasible investment opportunities, thereby moving towards the ultimate personal goal of financial independence.

(3)

List of tabies

TABLE 3.1: SALES TURNOVER CORRELATIONS 24 TABLE 3.2: SALES TURNOVER GROWTH CORRELATIONS 25

TABLE 3.3: AVERAGE SHARE PRICE CORRELATIONS 26 TABLE 3.4: AVERAGE SHARE PRICE GROWTH CORRELATIONS 27

TABLE 3.5: SALES TURNOVER REGRESSION MODELS 31 TABLE 3.6: SALES TURNOVER 11 FACTOR REGRESSION MODEL 32

TABLE 3.7: SALES TURNOVER MOST SIGNIFICANT REGRESSION MODEL 33

TABLE 3.8: SALES TURNOVER GROWTH REGRESSION MODELS 35 TABLE 3.9: SALES TURNOVER GROWTH 11 FACTOR REGRESSION MODEL 36

TABLE 3.10: SALES TURNOVER GROWTH MOST SIGNIFICANT REGRESSION MODEL 37

TABLE 3.11: AVERAGE SHARE PRICE REGRESSION MODELS 39 TABLE 3.12: AVERAGE SHARE PRICE 11 FACTOR REGRESSION MODEL 40

TABLE 3.13: AVERAGE SHARE PRICE MOST SIGNIFICANT REGRESSION MODEL 41

TABLE 3.14: AVERAGE SHARE PRICE GROWTH REGRESSION MODELS 43 TABLE 3.15: AVERAGE SHARE PRICE GROWTH 11 FACTOR REGRESSION MODEL 44

TABLE 3.16: AVERAGE SHARE PRICE MOST SIGNIFICANT REGRESSION MODEL 45

(6)

List of figures

FIGURE 1.1: GROWTH OPPORTUNITIES IN SHARE ACCELERATION AND REACCELERATION 2

FIGURE 2.1: DUPONT ANALYSIS 16 FIGURE 3.1: NORMALITY OF REGRESSION ERRORS 29

FIGURE 3.2: RESIDUAL SCATTER PLOT 30 FIGURE 3.3: SALES TURNOVER MOST SIGNIFICANT FACTOR HISTOGRAM 34

FIGURE 3.4: SALES TURNOVER GROWTH MOST SIGNIFICANT FACTOR HISTOGRAM 38 FIGURE 3.5: AVERAGE SHARE PRICE MOST SIGNIFICANT FACTOR HISTOGRAM 42 FIGURE 3.6: AVERAGE SHARE PRICE GROWTH MOST SIGNIFICANT FACTOR HISTOGRAM 46

(7)

CHAPTER 1 INTRODUCTION

1.1 BACKGROUND

Only six percent of South Africans can retire financially independent (FNB, 2004). Put differently, only six percent can afford to retire. An even more shocking statistic is that only nine percent of South Africans who are members of retirement funds will be financially independent once they reach retirement age (Van der Waldt & Van Heerden, 2007). It is therefore evident that individuals cannot rely on a primary retirement fund for eventual financial independence. Additional wealth building mechanisms need to be incorporated into retirement portfolios. The onus rests upon the individual to take ownership of and actively participate in the creation of personal wealth.

The stock market is cited to be one of the greatest tools ever invented for building wealth (Magliolo, 2005:XV). As such, stocks are part of nearly any investment portfolio. The Johannesburg Stock Exchange (JSE) thus offers South African investors the ideal opportunity for building long term wealth and eventually reaching financial independence.

According to Magliolo (2005:135), it is best to limit the number of shares in an investment portfolio to between 10 and 12, spread across at least 3 sectors. Magliolo stresses the importance of active portfolio management, and as such, a larger amount of shares is impractical to manage. The diversification level achieved with the recommended number of shares is adequate for the average investor. Considering the relative low recommended number of shares in the ideal portfolio, it logically follows that share selection is absolutely crucial to portfolio success.

Depending on the risk / return profile of an investor, a proportion of the investment portfolio shall consist of high growth stocks. Ideally, investors want to capitalise on the life cycle stages that organisations typically go

(8)

through. The typical corporate life cycle may be disclosed in five common

stages: birth, growth, maturity, revival and decline (Miller & Friesen,

1984:1161). The growth and revival stage of the corporate life cycle offer

investors the opportunity for growth in share value. This process is

graphically illustrated in Figure 1.1 (Bernstein, 2006).

Figure 1.1: Growth opportunities in share acceleration and

reacceleration

Growth

Opportunity:

Acceleration

i +

X

Growth Trap:

iecele? rtiorv

L ecline

Growth

Opportunity:

Sustain* bllit\

^-~¥

bf

C o r : - .■■■■: " : , ■ ' ':,■■

Source: Alliance Bernstein, 2006

1.2 PROBLEM STATEMENT

A significant problem for investors is to determine the optimum point of

investment when corporate life cycle is considered. It can be argued that

identifying the time of investment for small and medium sized companies is

relatively easy because these are "naturally" growing companies. None the

less, it is known that even determining the optimum points of investment in

growing companies are not always easy to identify. Identifying time of

investment for companies in the maturity and declining stages of life cycle

might even be more difficult. It is therefore important to try and develop

(9)

guidelines, based on leading indicators of future growth, which may assist in the consideration and selection of alternative investments.

1.3 OBJECTIVE

1.3.1 Main Objective

The main objective of this study is to research and if possible, identify and develop investment guidelines based on possible future growth indicators of organisations listed on the JSE.

1.3.2 Sub-objectives

The sub-objectives of this study are defined as follows:

• To investigate indicators that suggest future growth or the revitalisation of businesses in the growth, maturity or decline phases of the company life cycle.

• To test the validity of the proposed leading indicators on the identified research samples.

1.4 RESEARCH METHODOLOGY

The literature review primarily focuses on identifying leading indicators that may predict a future growth stage in an organisation. These leading indicators shall be tested on actual share price data in the empirical research part of the study to determine which of the identified indicators shall be the most valuable to the investor. Four dependent variables, signifying typical growth metrics, are used in the study:

• Sales turnover.

• Sales turnover growth. • Average share price; and • Average share price growth.

(10)

The research sample is compiled by identifying the top 100 shares by turnover on the JSE as in 2007. Data is extracted for 1992 to 2007 for the share sample, offering 16 years of research data.

The possible leading indicators identified in the literature review are statistically tested on the selected organisations, to ascertain whether any of the indicators are significantly related to any of the dependent variables. The possible leading indicators, or independent variables, are:

Numerous regression models are calculated to determine which of the possible leading indicators are statistically significant in explaining future growth metrics or dependent variables. The top 5 leading indicators are identified for each growth metric. Subsequently, the top 3 leading indicators are isolated.

1.5 SCOPE OF THE STUDY

Financial data of the top 100 shares by turnover, listed on the JSE, is statistically analyzed using multiple linear regression. The top 5 most relevant future growth indicators are identified for each of the 16 years of data being analyzed. The overall top 3 most significant possible future growth indicators are identified based on the number of occurrences in the yearly top 5 factors.

(11)

1.6 LIMITATIONS OF THE STUDY

The statistical analysis performed in this study is limited to multiple linear regression. The assumptions required for regression is tested only on a limited number of regression models due to the very large number of regression models generated. For the purposes of this study, it is assumed that the assumptions of regression are met for all the models generated. This assumption is based on a residual analysis performed on a limited number of regression models. More advanced studies, outside the scope of this study, can confirm that the assumptions of regression are indeed met for all the regression models generated in this study.

The performance metrics or dependent variables are analysed within years, meaning the "components" of a metric in a given year is determined. It is not attempted to relate the performance metrics of a specific year to previous year's data. Therefore, no "lag" analysis is performed.

This study only focuses on identifying the most relevant future growth indicators. The study does not focus on how these identified variables should be used for investment selection and due diligence.

1.7 LAYOUT OF THE STUDY

Chapter 1: Introduction

Chapter 2: Value Investment Theory Chapter 3: Value Investment Researched Chapter 4: Conclusion

References Appendixes

(12)

CHAPTER 2 VALUE INVESTMENT THEORY

2.1 INTRODUCTION

Value based management and the concept of corporate valuation are discussed briefly in the first section of the literate review. Corporate value drivers are derived from insights gained from the corporate valuation model. These value drivers are discussed in detail in the second section of the literature review. The popular DuPont analysis is also discussed. The last section of the literature review examines existing research into share price growth indicators.

2.2 VALUE BASED MANAGEMENT

It is often cited that the primary objective of management is to maximize the shareholder wealth of companies (Brigham & Ehrhardt, 2005:507). The concept of value based management is based on the following corporate valuation model: The value of an organisation is the present value of the expected future free cash flows, discounted at the weighted average cost of capital (WACC).

2.2.1 Corporate valuation

The value of a firm can thus be calculated as follows:

FCF, FCF2 FCF3 FCF„

Value- —-r + - —7^ + —7T- + ---- + _{(1 + WACC)' (1 + WACC)}

2 (1 + WACCf "" (1 + WACC)"

Where:

FCF = Free cash flow

WACC = Weighted Average Cost of Capital

Free cash flow (FCF) is defined as the cash available for distribution to investors after the company has made all the fixed asset and working capital investments required for sustaining ongoing operations. Thus, FCF is the

(13)

difference between the net operating profit after taxes and the net investment in operating capital, or:

FCF = NOPAT - Net investment in operating capital Where:

NOPAT = Net Operating Profit After Taxes

From the above equations, it is evident the value of a company is closely linked to the free cash flows and the cost of capital. The present value of a firm depends on the future cash flows for the firm's entire lifetime, presenting a complex calculation. Fortunately, assuming that a point of constant growth shall be reached after time N, the value of the firm at that point is defined as:

FCF

_N+\

op(atLimeN) y/J^QQ _ g

Where:

" o p (at lime N) = Corporate value at time N

FCF = Free cash flow g = Growth in sales

WACC = Weighted Average Cost of Capital

The free cash flows can however be disseminated further to expose the fundamental value drivers. The corporate value at time N can be rewritten in two general forms. The first form is:

v _{eip (al lime N)}= Capital^ +

Where: V0p (at time N) OP CR g WACC Capital^ SalesN Sales N(\ + g) WACC-g OP-WACC r CR^ } + 8

= Corporate value at time N

= Operating profit

= Capital Requirements = Operating Capital / Sales = Growth in sales

= Weighted Average Cost of Capital = Capital invested at time N

(14)

Therefore, the value of a corporation can be expressed as the capital provided by investors at time N, combined with the additional value created by management, equivalent to market value added (MVA).

The second form of the corporate valuation equation is as follows:

i CapitalN{EROICN-WACC)

^( a, am cN , = CapitalN + m c c _ "

Where:

Vop (at time N) = Corporate value at time N

EROICN = Expected return on invested capital at time N

g = Growth in sales

WACC = Weighted Average Cost of Capital CapitalN = Capital invested at time N

The second form also contains a component for invested capital and a component for MVA. This component for MVA depends on the mixture of expected return on invested capital (EROIC), and the weighted average cost of capital (WACC). It is evident that value shall be created if the expected rate of return is greater than WACC.

2.2.2 Value / wealth drivers

Considering the previous equations, the value of a firm can be expressed in terms of four fundamental wealth drivers (Brigham & Ehrhardt, 2005:518):

• Operating profit.

• Capital Requirements (Operating Capital / Sales). • Growth in sales; and

• Weighted Average Cost of Capital.

These drivers are discussed in detail in the following section.

The expected return on invested capital (EROIC) also forms an important metric in the corporate valuation model. EROIC is defined as the expected

(15)

NOPAT for the coming year divided by the amount of operating capital at the beginning of the year:

NOPAT\ EROICN = N+l

Capital N

Where:

E R O I C N = Expected return on invested capital at time N

NOPATN+1 = EBITN (1 - TaxRate) at time N+1

Capita^ = Capital invested at time N

E B I T N = Earnings before interest and taxes at time N

Therefore, the drivers of EROIC are the earnings before interest and taxes (EBIT) and the amount of capital invested in the organisation. The EBIT is closely related to the relationship between income and the expenditures required to generate that income. Therefore, the EBIT is closely related to sales turnover, profit margins and keeping expenditures under control.

2.2.3 Performance metrics

Weaver & Weston (2003:15) identify four alternative performance metrics that may be used to monitor the effectiveness of value based management efforts:

• Basic intrinsic value analysis (IVA).

• Discounted cash flow (DCF) methodology.

• Economic profit (EP) or Economic value added (EVA) approach; and • Market-to-book ratio or market value added (MVA) analysis.

Weaver & Weston find that the alternative performance metrics are highly correlated. It is also found that the standard financial ratio analysis as expressed in the DuPont formulation are also significantly related to the implementation of value based management and related market performance. The DuPont formulation shall be investigated further in this study.

(16)

2.2.4 Conclusion

Management can use the fundamental wealth drivers offered by the corporate

valuation model to guide strategic and operational decisions to ensure that

shareholder wealth is maximized. Thus, it can be argued that organisations

actively practising value based management offer the prospect of future

growth and wealth creation to shareholders.

2.3 CORPORATE VALUE DRIVERS

2.3.1 The cost of capital

The weighted average cost of capital (WACC) is listed in the previous section

as one of the fundamental wealth drivers. This section focuses on the core

components of and calculation of WACC. The optimum capital structure is

also discussed. WACC is determined by three major capital components:

debt, preferred shares and ordinary shares.

2.3.1.1 Cost of capital components

Cost of debt

The cost of debt is defined as the rate of return that debt holders require, r

d

.

The cost of debt of a firm is highly dependent on the type of firm and the

associated risk profile. The interest that is charged on debt is tax deductible,

thus the after tax cost of debt is applicable to the calculation of WACC.

Therefore,

Cost of debt (after tax) = r

d

(1 -T) where T is the firm's marginal tax rate.

Cost of preferred stock

The cost of preferred stock, r

ps

is calculated as the preferred dividend, D

ps

,

(17)

rp*

Where:

rps = Cost of preferred stock

DpS = Preferred dividend of stock

Pn = Net issuing price of stock

The dividends paid out on preferred stock are not tax deductible, and therefore no tax adjustment is made as with the cost of debt.

Cost of common stock

The cost of common stock, rs, is the rate of return that shareholders can

expect to earn on an equivalent risk investment. The cost of common stock can be approximated by using three typical methods:

• CAPM - the capital asset pricing model. • DCF - the discounted cash flow; and • Bond-yield-plus-risk-premium.

2.3.1.2 WACC

The weighted average cost of capital is calculated as the sum of the cost of debt, the cost of preferred stock and the cost of common stock:

WACC = wdrd (l - T) + wp,rps + WLJS

Weight of debt in the corporate capital structure Pre-tax cost of debt

Marginal tax rate

Weight of preferred stock in the corporate capital structure Cost of preferred stock

Weight of common equity in the corporate capital structure Cost of common equity

(18)

The cost of capital may be influenced by the policies adapted by an organisation. Brigham and Ehrhardt (2005:323) offer three factors that a firm can control to affect the cost of capital:

• Capital structure policy. • Dividend policy; and • Investment policy.

The first factor, the capital structure policy, refers to the combination of debt, preferred stock and ordinary stock used for financing an organisation. Each firm has an ideal capital structure that minimizes WACC. Since WACC features in the denominator of the corporate valuation equation, a low WACC shall result in a high value, ceteris paribus. This is confirmed by Black, Wright and Davies (2001:9), who propose that shareholder value is produced when the equity returns of a company exceed the cost of that capital (WACC). It can be argued that an organisation not operating at the ideal capital structure, but migrating towards it, can offer value to the investor.

The second factor, the dividend policy, may affect the rate of return required by investors, rs. This shall in turn impact directly on the computation of

WACC.

The last factor, the investment policy, refers to the assumption that new capital will be invested in assets with a similar risk profile similar to existing assets. If, however, capital is invested in an entirely different line of business, the new cost of capital shall encapsulate the risk involved in the new venture.

2.3.2 Growth in sales - g

The expected future growth of sales generally has a positive effect on a firms value, provided that the company is profitable enough (Brigham and Ehrhardt, 2005:518). According to Nardinelli (2002:10), the expected resultant future earnings represent the growth engine of the firm's value, fuelling the firm's future stock price.

(19)

The relationship between future earnings and share price has been the topic of numerous research studies. Beaver, McAnally and Stinson (1997:139) identified earnings changes as a major contributor to share price change, using a simultaneous equations approach.

Predicting the growth in sales is a difficult task, since unknown future factors have to be considered. From an investor's point of view, it may be argued that the growth as such is less important than the growth relative to the industry index in which the firm operates. Generally, significant future events or circumstances grossly affecting future expectations shall have a similar effect on all the firms operating in that industry.

Growth can be estimated, under strict assumptions, using the Retention Growth Model (Brigham and Ehrhardt, 2005:318) as:

g = ROE (Retention Ratio) Where:

g = Estimated growth ROE = Return on equity

The above equation is only valid if the following assumptions are made: • Constant payout rate.

• Constant ROE for new investments.

• No new shares are issued or issued at book value; and

• Future projects do not significantly alter the risk profile of the company.

Therefore, in general terms, investment potential is directly linked to the expected future growth in sales. The exception occurs when this growth is capital intensive at a high cost of capital (Brigham and Ehrhardt, 2005:519).

(20)

2.3.3 Operating Profitability-OP

The operating profitability of a firm is defined as the after tax profit per monetary unit of sales. The operating profitability shall always have a positive effect on the valuation of a firm. Thus, if improvements are expected in the operating profitability of a firm, the value of the firm can be expected to increase. This increased value shall be reflected in firm's share price. Interestingly, research suggests that share price changes reflect information earlier than earnings do (Beaver, Lambert & Ryan, 1987:154). This theory is closely related to the strong form of the efficient market hypothesis, stating that all information, past, present and future, is reflected in the share price of a firm (Fama, E. 1963:420 & Samuelson, P. 1965:9). Therefore, the anticipated increase in operating profit and resultant change in value are already priced in to the share price before the earnings are actually realised.

A firm's profit margin is often derived from the firm's pricing power resulting from factors such as product positioning, product innovation, brand name recognition, first mover advantage and market niches (Soliman, MT. 2007:3).

The operating profit also features as one of the three fundamental elements in the calculation of return on equity (ROE) as defined in the extended DuPont equation. The DuPont system is discussed in more depth in a section dedicated to the use of DuPont to increase the return on equity of an organisation.

2.3.4 Capital Requirements - C R

The capital requirements ratio measures the amount of operating capital per unit of sales generated. A lower capital requirements ratio is beneficial to an organisation as less capital is required to generate sales and future profits. The capital requirements of start-up organisations shall typically be higher due to the initial high outlay of capital required for productive capacity. As an organisation matures, the capital requirements ratio should lower due to higher process efficiencies and process maturity. An increasing capital

(21)

requirement ratio may indicate a decrease in the future productivity, unless new products innovation or research is responsible for the more capital intensive operations.

2.4 DUPONT ANALYSIS

DuPont analysis is a popular methodology for financial statement analysis. The return on net operating assets is decomposed into two multiplicative components: profit margin and total asset turnover. Return on equity is defined as the return on net operating assets multiplied by the equity multiplier. Monteiro (2006:3) states that ROE is perhaps the most important ratio an investor should consider. It should be noted that ROE is affected by the gearing levels of a company, and as such, investors have to interpret ROE with caution (De Wet & Du Toit, 2006).

The extended calculation equation of the DuPont is as follows:

ROE = (Profit Margin) x (Total Asset Turnover) x (Equity Multiplier)

Net income Sales Total Assets

— x x Sales Total Assets Common Equity

The components of ROE can be expressed graphically in Figure 2.1 (Correia, Flynn, Uliana and Wormald, 2003:5-20).

(22)

Figure 2.1: DuPont Analysis Net Profit Margin Total Asset Turnover Total Assets Ordinary Equity

Source: Correia, Flynn, Uliana and Wormald, 2003

DuPont offers management three key, tangible performance areas, namely profit margin, total asset turnover and the equity multiplier. Profit margin is closely related to product positioning. Changes in a firm's profit margin measures the growth rate of operating income relative to the growth rate of sales. Total asset turnover measures asset utilization and efficiency. The equity multiplier refers to the use of the correct debt / equity financing mix to minimize the effective cost of capital. Changes in the total asset turnover reflect a change in the firm's productivity, measuring the growth in sales relative to the growth in operating assets.

Research shows that changes in the components of DuPont, and not the components as such, have explanatory power with respect to changes in the future profitability of a firm {Fairfield & Yohn; 2001:371). Fairfaild and Yohn found that changes in the total asset turnover offer the most insight into future returns. Research by Soliman (2004:5) also found that abnormal return on net operating assets due to abnormal total asset turnover is more persistent than abnormal return that is due to abnormal profit margins. This conclusion is consistent with an intuitive approach, since increased asset utilization should

(23)

offer future benefit to the organisation, being reflected in increased future profitability. Rappaport (1986:43) however warns that total asset turnover is highly influenced by inflation, and as such should be interpreted with caution.

Soliman (2004:3) suggests that although the return on net operating assets may revert to economy wide benchmarks, the components as indicated by the DuPont approach will not revert to economy wide benchmarks. It is suggested that the components shall typically cluster according to industry. The fact that the return on net operating assets may revert to economy wide benchmarks is strongly backed by the following espouse (Stigler, 1963):

"There is no more important proposition in economic theory than that, under competition, the rate of return on investment tends toward equality in all industries."

Research by White, Sondhi, and Fried (1998:190) indicate that a strong negative relationship exists between profit margin and asset turnover, implying that although similar levels of return on net operating assets are achieved across industries, this returns are the result of different combinations of profit margin and asset turnover. Soliman (2004:4) suggests that the profit margin and asset turnover are generally more informative when considered relative to industry competitors, since vast differences may exist between different industries.

It can therefore be concluded that profit margin and net asset turnover should not be viewed in isolation when comparing organisations, but according to industry sector. It shall therefore be useful to normalize the previously mentioned factors according to industry sector, thereby allowing cross-sector comparison. On the other hand, return on total net operating assets may be compared directly across different industry sectors, since ROA depends mostly on economic factors.

(24)

2.5 GROWTH INDICATOR RESEARCH

2.5.1 Dynamic Gap Investing - Bernstein/Alliance

Bernstein/Alliance (2006:10) has developed a strategy labelled "dynamic gap investing" that attempts to identify shares that offer high growth potential. This strategy is briefly discussed and may contribute to the empirical research part of this study. "Surprisers" are identified as stocks that offer above average returns to investors. The research indicates that these "surprising" shares have the following common traits:

• Positive earnings growth. • Increasing growth rate; and

• Accelerating pace of increased growth.

Bernstein/Alliance offer five guidelines for selecting growth stock winners, listed in order of predictive power:

• An earnings growth rate that is currently increasing faster than previously.

• Positive stock-price momentum over the past 12 months. • Year-over-year earnings growth.

• Increasing returns on equity; and • Increasing free-cash-flow margins.

The above future growth indicators shall be tested in the empirical part of this study for validity and usefulness.

2.5.2 Fundamental analysis by Nguyen

Nguyen (2003:5) constructs a share selection scorecard based on a combination of financial ratios designed to capture short-term variations in the company's operations, profitability and financial policy. Although closely related to system proposed by Abarbanell and Bushee (1998:19), Nguyen focuses more on variables that are related to capital structure, dividend yield

(25)

changes and operations, thereby providing signals of the company's future prospects. Nguyen includes the following variables:

• Profitability

o Return on assets (ROA).

o Change in return on assets (AROA); and o Change in return on equity (AROE). • Operating efficiency

o Change in turnover. o Change in inventory; and o Change in inventory turnover. • Leverage and dividends

o Change in leverage; and o Change in dividend yield.

Nguyen found that fundamental analysis is helpful in predicting future stock returns. It was found that the high score stocks achieved a remarkable return of 1 1 % above market average, offering credibility to the use of his accounting variables as possible future growth indicators.

Jegadeesh (1990:881) shows that share returns tend to exhibit short-term momentum. It can be argued that this short term momentum contributes to the suitability of using the change in variables. For example, if the ROE increased for the past few periods, it can be argued that the chances are good for another period of increased ROE. The momentum concept shall be tested in the empirical part of this study.

2.5.3 Long-Term Return Reversals

DeBondt and Thaler (1985:557) identify share "losers" as shares that have had poor returns over the past three to five years. "Winners" are defined as share returns over a similar period. Surprisingly, the main result of DeBondt and Thaler's research is that losers have much higher average returns than

(26)

winners over the next three to five years! The empirical part of this study shall test the above.

2.6 CONCLUSION

Several possible leading indicators have been identified in the literature review. These indicators shall be investigated for predictability power through testing on actual historical share performance data of organisations listed on the Johannesburg Stock Exchange in the following part of this study.

(27)

CHAPTER 3 VALUE INVESTMENT RESEARCHED

3.1 INTRODUCTION

This chapter contains the empirical research part of this study. The first section elaborates on the selection of growth metrics and possible growth indicators. Thereafter, the selection of a suitable sample is discussed as well as the sampling period. Correlation analysis per growth metric is presented, followed by stepwise regression models. The top 5 significant indicators are determined for each metric over the sampling period. Thereafter, the overall top 3 significant indicators are extracted and discussed.

3.2 INDICATOR SELECTION

Several possible leading indicators for future growth have been identified in Chapter 2 of this study. These indicators form the independent variables in the statistical analysis conducted and described in this Chapter. The measurement of future growth and growth potential is encapsulated by the dependent variables.

3.2.1 Dependent variables

The dependent variables used in the statistical analysis, representing typical growth metrics, are as follows:

• Sales turnover.

• Sales turnover growth. • Average share price; and • Average share price growth.

(28)

3.2.2 Independent variables

The possible future growth indicators, as identified in Chapter 2, are as follows:

3.3 SAMPLE SELECTION

The statistical sample is selected as the top 100 companies listed on the Johannesburg stock exchange ranked by sales turnover in 2007. It is opted to select the sample as such to ensure that no sampling bias errors are encountered. Refer to Appendix A for the selected companies and the respective sales turnover figures for 2007.

3.4 SAMPLING PERIOD

Annual historical data have been collected for the sample JSE companies for the years 1992 through to 2007, forming a dataset consisting of 16 years.

(29)

3.5 CORRELATION ANALYSIS

The correlations between all the dependent and independent variables are calculated to ensure that independent variables are not highly correlated. A correlation matrix is calculated for each year from 1992 to 2007. The 16 off correlation matrixes are combined into a single matrix by calculating a matrix of mean correlation coefficients. This combined correlation matrix is purely intended to assist in interpreting the vast amount of data and is included in Appendix B.

It is evident that there exists a significant correlation between the financial leverage ratio and the return on equity (ROE) indicator. This correlation averages to 67.8%. It is therefore decided to remove the financial leverage ratio from any further statistical analysis to ensure that the independent variables are significantly independent.

The correlation of each dependent variable with the set of independent variables are analysed to identify any significant relationships that will contribute to the outcome of this study.

3.5.1 Sales turnover

The correlation of sales turnover with the various possible growth indicators are shown in Table 3.1 for 1992 to 2007. The correlation coefficient for each independent variable with sales turnover is calculated for each of the 16 years as well as the mean correlation coefficient, shown in Table 3.1. The mean correlation coefficients have no statistical significance other than providing a metric for comparing the relative significance of the various independent variables.

(30)

Table 3.1: Sales turnover correlations .£ "5) i_ re E "5 i_ Q . O) C re b . CD Q . O re EC ■jfl c ED E eu i_ 3 O" at c *5) i_ re S o i_ Q . a> Z < O CC Tota l asse t turnove r CO Q, LU i_ re .c in i_ a 5 o j = (A ra o Inventor y ■o c a> ■a '> a 1 9 9 2 -0.066 -0.089 -0.046 0.032 -0.063 -0.020 0.003 0.228 0.431 0.973 -0.231 1 9 9 3 -0.047 -0.093 -0.055 -0.026 -0.023 -0.042 -0.019 0.214 0.406 0.970 -0.127 1 9 9 4 -0.049 -0.049 -0.070 -0.158 -0.013 0.000 -0.098 0.177 0.281 0.931 -0.251 1995 -0.031 -0.027 0.088 0.026 0.005 -0.075 -0.054 0.149 0.247 0.958 -0.197 1996 -0.056 -0.014 -0.070 0.031 0.004 -0.089 -0.070 0.077 0.309 0.945 -0.095 1997 -0.061 -0.072 -0.054 -0.064 -0.010 0.043 -0.116 0.223 0.354 0.949 -0.074 1998 0.045 0.077 0.068 -0.015 0.063 -0.053 -0.192 0.428 0.449 0.926 -0.024 1999 -0.129 0.082 -0.043 0.015 -0.045 -0.190 -0.126 0.245 0.212 0.883 0.070 2 0 0 0 -0.098 -0.088 0.071 -0.059 -0.089 -0.082 -0.044 0.097 0.165 0.811 -0.128 2001 -0.078 -0.088 0.157 0.014 0.088 -0.025 -0.067 0.205 0.303 0.823 -0.092 2 0 0 2 0.025 -0.052 0.015 0.047 0.076 -0.003 -0.064 0.180 0.247 0.857 -0.036 2 0 0 3 -0.045 0.009 0.113 -0.111 0.083 -0.024 -0.018 0.171 0.076 0.841 0.004 2 0 0 4 -0.051 0.006 0.093 -0.051 0.094 -0.029 -0.041 0.150 -0.002 0.837 -0.129 2 0 0 5 -0.108 -0.036 0.144 0.009 0.101 -0.092 -0.044 0.153 -0.071 0.788 -0.045 2 0 0 6 -0.077 -0.083 0.108 -0.034 0.056 -0.023 0.050 0.205 -0.035 0.824 -0.005 2007 -0.053 -0.089 0.050 0.048 0.019 0.015 0.083 0.270 0.291 0.829 0.079 Mean -0.055 -0.038 0.036 -0.019 0.022 -0.043 -0.051 0.198 0.229 0.884 -0.080

The mean correlation of 88.4% between sales turnover and inventory, ranging from 78.8% to 97.3%, confirms the logic conclusion that higher sales requires higher inventory. The mean correlation between sales turnover and cash flow per share is calculated at 22.9% ranging from - 7 . 1 % in 2005 to 44.9% in 1998. Similarly the correlation of earnings per share and the sales turnover is 33.9% with a range of 7.7% to 42.8%. Therefore, inventory, cash flow per share and earning per share appear to be the most relevant factors when considering sales turnover. The regression model section of this study shall confirm the relevance of these three factors. Considering that the correlation between sales turnover and the remaining independent variables range between 1.9% and 8%, no obvious, useful inferences can be made with regard to the usefulness of these variables to predict sales turnover.

(31)

3.5.2 Sales turnover growth

The correlation between sales turnover growth and the possible growth indicators are represented by the correlation coefficients shown in Table 3.2. The mean correlations calculated for the entire period does not present any useful inferences to be made due to the low correlation coefficients.

Table 3.2: Sales turnover growth correlations

c O) l _ CO S o l _ Q . o> c CO l _ 0> a. o o n « c re S o ■_ Q . z

<

O CC 0) > o c 3 o 3 1 o O . LU c •a 0) C Q 1992 -0.151 -0.109 -0.023 -0.035 -0.129 -0.046 -0.036 -0.147 -0.155 -0.102 0.000 1993 -0.188 -0.138 -0.116 -0.446 -0.120 -0.225 -0.056 0.216 0.006 -0.093 -0.046 1994 -0.457 -0.166 0.032 0.074 -0.477 -0.662 0.147 -0.026 -0.060 -0.170 -0.104 1995 0.091 -0.106 0.024 0.034 0.006 0.106 -0.021 0.079 0.025 0.081 -0.064 1996 -0.157 -0.084 -0.121 0.048 -0.154 -0.161 0.092 -0.039 -0.157 -0.183 0.037 1997 -0.078 -0.070 -0.038 0.043 -0.028 0.001 -0.114 0.026 -0.096 -0.150 -0.128 1998 0.000 -0.120 0.052 0.011 0.064 0.019 -0.091 0.658 0.210 0.559 -0.058 1999 0.011 -0.345 -0.006 0.059 -0.023 -0.002 -0.064 -0.056 0.002 -0.053 -0.104 2000 0.030 0.017 0.272 -0.023 0.018 0.469 0.190 0.000 0.031 -0.093 -0.030 2001 -0.236 -0.244 0.210 -0.039 0.244 0.221 0.176 0.238 0.131 0.249 0.115 2002 0.312 -0.043 -0.019 0.126 -0.069 -0.071 -0.135 -0.056 -0.103 0.146 -0.080 2003 0.262 0.078 0.063 -0.051 -0.052 -0.001 -0.084 -0.074 -0.026 -0.040 -0.082 2004 0.300 -0.093 -0.334 0.038 -0.112 0.158 -0.128 -0.065 0.029 -0.095 -0.081 2005 0.052 -0.098 -0.043 0.151 -0.022 0.056 -0.034 -0.017 -0.074 -0.026 -0.190 2006 0.205 -0.393 0.070 -0-077 0.035 0.283 0.133 0.175 0.152 0.134 -0.021 2007 0.185 -0.314 -0.281 0.052 -0.166 0.120 -0.045 -0.059 -0.062 -0.009 -0.121 Mean 0.006 -0.141 -0.019 -0.005 -0.066 0.011 -0.003 0.052 -0.011 0.005 -0.060

3.5.3 Average share price

Refer to Table 3.3 for the correlation between the average share price and the independent variables. The most significant correlation relationship exists between share price and earnings per share with a correlation of 78.8% ranging from 52% in 2000 to 93.8% in 1997. The next highest correlation is that of cash flow per share and average share price at 63.1%. These findings

(32)

represent two popular methods of share selection, looking at the cash flow per share and the earnings per share.

Table 3.3: Average share price correlations

c "5> CD S o b . O. D> C CD b . a> a . O o n ec CA c a> E fi 5 a> cc 2 '5. ra U LU o EC o n CC c o c a> "3 DC c "o> b . CD S o b . a. a) Z < O DC b . a> > o c b . a> <A <A «s o (A Q . LU a> b . C3 CA b . CD a. o (A ra O o c > ■D CD >-■o c a> ■o "> 5 1992 0.522 0.598 0.117 -0.286 0.535 0.055 -0.284 0.805 0.754 0.136 -0.325 1993 0.249 0.437 0.100 -0.209 0.282 -0.011 -0.304 0.853 0.759 0.132 -0.317 1994 0.356 0.430 0.119 -0.406 0.403 0.063 -0.299 0.873 0.798 0.174 -0.188 1995 0.486 0.654 0.003 -0.060 0.550 0.000 -0.373 0.822 0.797 0.088 -0.073 1996 0.390 0.571 0.006 -0.202 0.492 -0.055 -0.378 0.839 0.883 0.199 0.006 1997 0.389 0.343 0.073 -0.161 0.529 0.047 -0.347 0.938 0.789 0.199 -0.035 1998 0.183 0.315 0.060 -0.163 0.224 -0.064 -0.315 0.821 0.850 0.461 -0.011 1999 0.107 0.111 0.106 -0.049 0.197 0.108 -0.257 0.658 0.593 0.280 0.234 2000 -0.024 -0.056 0.229 0.164 -0.055 0.171 -0.249 0.520 0.678 0.295 0.132 2001 0.052 -0.050 0.709 -0.019 0.052 0.419 -0.132 0.827 0.865 0.179 0.048 2002 0.381 -0.017 -0.134 -0.137 0.221 0.442 -0.052 0.819 0.849 0.182 -0.015 2003 0.263 -0.030 0.100 -0.154 0.217 0.334 -0.106 0.842 0.387 0.214 -0.008 2004 0.127 -0.016 -0.014 -0.186 0.138 0.160 -0.128 0.565 0.164 0.269 0.066 2005 0.163 0.065 -0.058 0.025 0.337 0.171 -0.153 0.892 0.065 0.335 -0.015 2006 0.197 0.000 0.010 -0.018 0.251 0.299 -0.072 0.818 0.124 0.389 0.058 2007 0.344 0.034 -0.031 0.068 0.320 0.489 -0.049 0.715 0.748 0.454 0.093 Mean 0.262 0.212 0.087 -0.112 0.293 0.164 -0.219 0.788 0.631 0.249 -0.022

3.5.4 Average share price growth

The correlation between the average share price growth, measured year on year, and the various independent factors are shown in Table 3.4. The most significant correlation, although only 16.1% on average, is the correlation between average share price growth and dividend yield. Considering that all the correlations calculated are relatively low, no useful inferences can be made regarding the "components" of average share price.

(33)

Table 3.4: Average share price growth correlations c o> CO £ o CL O ) c "5 01 a o g cc at c Q) E k. CT a: £ '5. ca u w O DC o *£ cc c .2 c £ "5 cc c "5> CO £ o CL a* z < O > o c 3 "5 CA ( A CO "(5 o 1 -( 0 CL UJ £ CO ■C ca a> a . £ o en CO O o c CU > c ■a a> > ■a c CD ■ a > 5 1992 0.037 -0.058 -0.107 -0.028 0.001 -0.128 -0.207 0.033 0.088 0.039 -0.471 1993 0.032 -0.090 0.010 -0.093 0.106 0.007 -0.065 0.106 0.034 -0.151 -0.289 1994 -0.019 -0.026 0.058 0.058 0.061 0.090 0.243 -0.004 -0.031 0.016 -0.024 1995 0.048 0.028 0.107 0.075 0.098 0.084 0.160 0.041 0.002 -0.112 -0.063 1996 -0.084 -0.090 0.028 0.089 -0.065 -0.057 0.349 -0.033 -0.047 -0.107 -0.134 1997 0.205 0.494 0.034 0.036 0.362 0.030 -0.263 0.062 -0.002 -0.074 -0.297 1998 -0.103 0.006 0.294 -0.115 0.113 0.083 -0.002 -0.029 -0.093 -0.146 -0.436 1999 0.118 ^0.410 -0.090 0.045 0.028 0.115 -0.087 0.001 0.040 -0.030 -0.198 2000 -0.099 -0.119 0.157 -0.013 -0.119 0.109 -0.085 0.223 0.201 0.172 -0.089 2001 0.058 0.019 0.158 0.022 -0.018 0.249 -0.058 0.115 0.155 0.055 0.115 2002 0.051 -0.150 0.100 -0.008 0.083 -0.001 0.134 0.126 0.116 0.023 -0.306 2003 0.395 -0.134 -0.291 -0.053 0.005 0.065 0.199 -0.025 -0.001 -0.175 -0.100 2004 -0.006 0.011 0.018 0.063 0.019 -0.026 0.015 -0.012 -0.008 -0.132 -0.071 2005 0.125 0.010 0.008 -0.048 0.132 0.072 0.089 0.038 0.060 -0.087 0.013 2006 0.216 -0.119 -0.032 -0.080 0.119 0.296 0.126 0.113 0.118 0.105 -0.019 2007 0.179 -0.056 -0.008 0.037 0.138 0.272 0.009 0.027 0.085 -0.101 -0.204 Mean 0.072 -0.043 0.028 -0.001 0.066 0.079 0.035 0.049 0.045 -0.044 -0.161

3.6 REGRESSION MODEL ANALYSIS

Multiple linear regression models are formulated for each growth metric in terms of the possible future growth indicators. A regression model is calculated for each year between 1992 and 2007. Furthermore, each model is refined a number of times by excluding the least significant variable. This exclusion process is repeated until all included variables are deemed as statistically significant. The first part of this section contains a discussion regarding the assumptions required for regression. The following four parts discuss the regression models formulated for each of the four dependent variables. Each section starts by summarizing the stepwise regression models and the usefulness or "fit" of the independent variables as related to the specific dependent variable. The "fit" is represented by the calculated adjusted coefficient of determination or adjusted R2 of the regression model.

(34)

Thereafter an 11 factor model and a most significant factor model are presented for each year of the study. A 95% level of significance is utilized to determine the most significant factors for each year. Standardized coefficients are calculated for each independent variable. The relative magnitude of each standardized coefficient represents the relevancy of the specific variable in explaining the dependent variable. The mean model fit and mean standardized coefficients are calculated to summarize the 16 years of data into an easily interpretable metric. Subsequently, a histogram is presented to identify the top 5 variables that are the most relevant to explaining the specific dependent variable. The last section combines the results obtained for each independent variable into a single histogram highlighting the most relevant growth metrics.

3.6.1 Assumptions

The three assumptions of regression (Levine et al, 2005:527) are as follows: • Normality of error.

• Homoscedasticity; and • Independence of errors.

The assumptions of regression are a prerequisite for the validity of any conclusions reached from the regression modelling. Due to the vast number of regression models generated, it is not practical for the scope of this study to present the assumptions of regression on each regression model. In the first step of the backward regression process, 64 regression models (16 years x 4 dependent variables) have been generated. Thereafter up to 11 backward regression models have been generated for each of the 64 base cases. For the scope of this study, a single regression model is analysed to ensure that the assumptions of regression are met.

The first assumption requires that the error around the line of regression be normally distributed. The regression standardized residual histogram for a regression model calculated for sales turnover is shown in Figure 3.1. It is

(35)

evident that there exists a strong tendency to normality. There are however a

few outlying data points up to 6 standard deviations from the normal. It is

therefore assumed for the purposes of this study that the normality

assumption is met.

Figure 3.1: Normality of regression errors

Histogram

Dependent Variable: Sales Turnover

Mean=-1.84E-16 Sld.Dev. =0.972

N=3Q

0 2 4 Regression Standardized Residual

The second assumption of regression is the requirement of homoscedasticity

that requires the variation around the line of regression to be constant. Figure

3.2 shows the regression standardized residuals plotted against the predicted

standardized regression values. It is evident from the pattern in the data

points that the residuals are mostly scattered around the zero line on the

regression standardized residual axis for the various values of the predicted

value axis. Therefore, the assumption of homoscedasticity is met since the

errors vary about the same for different values on the predicted value axis.

(36)

Figure 3.2: Residual Scatter plot

Scatterplot

Dependent Variable: Sales Turnover

Regression Standardized Predicted Value

The last assumption is the independence of errors in the regression model. It is evident from Figure 3.2 that no identifiable pattern exists in the scatter plot. The distribution appears to be random. Therefore, the assumption of independence of errors is met.

For the purposes of this study, it is assumed that the assumptions of regression are met. This assumption is based on a residual analysis performed on a single regression model. More advanced studies, outside the scope of this study, can confirm that the assumptions of regression are indeed met for all the regression models generated in this study.

Sales turnover

Table 3.5 contains a summary of the stepwise regression models for each year of the study for sales turnover. The adjusted coefficient of determination or adjusted R2 of each regression model is shown in the table. As an

(37)

adjusted model fit is 85.9%. When only factor 1 is considered, the model fit is still 86%. In general, when 11 variables are considered the adjusted model fit range between 66.0% in the 2005 11 factor model and 96.2% in the 1993 11 factor model. Sales turnover can be significantly explained by only 1 variable,

inventory, in 3 of the years considered, 1994, 1997 and 1999 as is shown in Table 3.7. It is therefore evident that a number of the variables considered are not statistically significant in predicting sales turnover. The next section shall look at which variables are significant.

Table 3.5: Sales turnover regression models

Year 11 Factor 10 Factor 9 Factor 8 Factor 7 Factor 6 Factor 5 Factor 4 Factor 3 Factor 2 Factor 1 Factor 1992 0.915 0.915 0.917 0.918 0.920 0.921 0.922 0.922 0.924 1993 0.962 0.963 0.964 0.965 0.965 0.966 0.965 0.964 0.964 1994 0.859 0.862 0.865 0.868 0.866 0.865 0.861 0.861 0.863 0.863 0.860 1995 0.946 0.947 0.947 0.947 0.946 1996 0.917 0.919 0.919 0.920 0.919 0.916 1997 0.896 0.899 0.901 0.903 0.902 0.900 0.901 0.902 0.901 0.898 0.895 1998 0.876 0.878 0.880 0.882 0.882 0.882 0.882 0.884 0.885 0.884 1999 0.767 0.772 0.777 0.781 0.784 0.788 0.790 0.786 0.786 0.782 0.780 2000 0.696 0.702 0.708 0.708 0.704 0.702 2001 0.780 0.784 0.786 0.789 0.790 0.791 2002 0.826 0.829 0.832 0.834 0.836 0.837 0.839 0.841 0.840 2003 0.822 0.825 0.827 0.830 0.831 0.831 0.832 0.828 0.828 2004 0.768 0.772 0.773 2005 0.660 0.664 0.666 0.667 0.664 0.664 0.668 2006 0.729 0.733 0.734 0.733 0.736 2007 0.701 0.706 0.710 0.713 0.717 0.720

Refer to Table 3.6 for regression models using all eleven independent variables to explain sales turnover for 1992 to 2007. The table contains the standardized coefficients of each regression model, highlighting the relative significance of each variable. A multiple linear 11 factor regression model is formulated for each year, resulting in a mean model fit or adjusted coefficient of determination of 82%, ranging between 66% in 2005 and 96.2% in 1993. Thus, the models are consistent in accurately predicting the sales turnover. It is evident that inventory is the most significant variable related to sales turnover, with a mean standardized coefficient of 0.869. This finding corresponds to the correlation analysis of the previous section, confirming the

(38)

obvious that higher sales volume usually requires greater inventory levels.

This observation may be of little use to the investor.

Table 3.6: Sales turnover 11 factor regression model

Standardized Coefficients I w co CD > ■ (O 3 t X (/) CC T3 3

m

c O S i_ CO

s

o a c to k. a> Q. o o CO CC CO +-• c 0) E 0) o-DC "co a . CB o LU O CC g CO CC c o c Q> +•* a> CC c '5> k. CO o k. Q_ 0) z < O DC a> > o c 3 CD CO CO CO "co *-o (7) CL HI a> Urn CO .c CO a> a o .c t o CO o L-o * ■ • c 09 > c 2 CD > I ■a c a> ■o I > 5 J1992 0.915 -0.069 -0.052 -0.026 - 0 . 0 2 3 0.117 - 0 . 0 2 2 0.083 0.070 -0.018 0.964 -0.032 1993 0.962 -0.104 -0.001 -0.042 - 0 . 1 5 2 0.126 0 . 0 2 7 0.078 0.039 -0.024 1.016 -0.035 1994 0.859 0.019 -0.399 0.006 - 0 . 0 7 2 0.360 - 0 . 1 5 7 0.144 0.163 -0.056 0.922 -0.095 1995 0.946 0.211 -0.806 0.120 - 0 . 0 5 4 0.595 - 0 . 2 7 7 0.161 0.184 0.041 0.956 -0.074 1996 0.917 0.772 -0.205 -0.104 0 . 0 4 6 -0.533 0 . 0 7 7 0.152 -0.015 0.143 0.950 -0.055 1997 0.896 -0.107 -0.322 0.029 - 0 . 0 5 9 0.272 - 0 . 1 2 3 0.065 0.340 -0.160 0.977 -0.008 1998 0.876 0.089 -0.052 -0.116 - 0 . 0 3 1 0.071 - 0 . 1 8 1 0.063 0.204 -0.077 0.912 -0.046 1999 0.767 -0.106 -0.039 -0.009 - 0 . 1 1 0 0.049 - 0 . 0 5 6 0.045 0.105 0.002 0.874 -0.132 2000 0.696 0.424 -0.181 0.188 0.008 -0.211 - 0 . 4 6 4 0.162 0.138 -0.018 0.773 -0.139 2001 0.780 0.220 -0.116 0.129 - 0 . 0 6 3 0.094 - 0 . 3 3 0 0.191 -0.058 0.148 0.815 -0.075 2002 0.826 0.151 -0.068 -0.161 0 . 0 1 2 0.121 - 0 . 3 3 2 0.217 -0.114 0.195 0.871 0.034 2003 0.822 0.172 -0.146 0.080 - 0 . 0 3 0 0.000 - 0 . 2 2 8 0.144 0.147 -0.077 0.901 -0.031 2004 0.768 0.472 0.084 0.156 - 0 . 0 0 8 -0.265 - 0 . 1 1 3 0.241 -0.391 0.483 0.755 -0.130 2005 0.660 0.395 -0.332 0.062 - 0 . 0 5 8 0.336 - 0 . 4 9 5 0.242 -0.304 0.258 0.740 -0.101 2006 0.729 0.581 -0.479 0.130 0.009 0.745 - 1 . 0 8 0 0.275 -0.170 0.156 0.779 -0.113 |2007 0.701 0.551 -0.183 -0.031 - 0 . 0 1 9 -0.079 - 0 . 5 4 5 0.297 0.070 0.206 0.695 -0.008| JMean 0.820 0.229 -0.206 0.026 - 0 . 0 3 8 0.112 - 0 . 2 7 8 0.160 0.025 0.075 0.869 -0.065

Table 3.7 contains the various most significant variable regression models

that are calculated for each year of the study. The relevant standardized

coefficients are listed in the column corresponding to that independent

variable.

(39)

Table 3.7: Sales turnover most significant regression model Standardized Coefficients k. CO 0) > 2 re 3 0-cc T3 S w 3 < c D) k. CO

s

0 k. O . O) c (0 k. Q. O 0 re DC (O 4-1 c E £ 5 0-a> DC re Q. re O UJ O DC 0 re DC c 0 '+* c a> a> DC c "5) k. re £ *-• •?■ 0 k. -*-> <D Z < O DC k. > 0 c k> 3 CD (A (A re 75 0 a. _LU <D k. re n (A k. a> a . 3S 0 •#— .c O) re O k. 0 *-• c 0) > c 2 > c a> ■g > 5 |1992 .924 .182 .103 .909 1993 .964 -.146 .054 1.010 1994 .860 .929 1995 .946 -.852 .154 .809 -.280 .132 .240 .962 1996 .916 .498 -.102 -.447 .142 .094 .969 1997 .895 .947 1998 .884 .172 .877 1999 .780 .885 2000 .702 .408 -.279 .270 -.292 -.472 .790 2001 .791 .392 -.142 .147 -.401 .187 .828 2002 .840 .160 .104 .890 2003 .828 .113 .097 .913 2004 .773 .383 .124 .188 • ; 'e .203 -.404 .494 .761 -.131 2005 .668 .577 -.255 -.449 .246 .780 2006 .736 .702 -.462 .129 .579 -1.169 .268 .799 |2007 .720 .440 -.166 -.493 .280 .243 .707 [Mean 0.827 0.486 -0.290 0.129 -0.146 0.075 -0.544 0.177 0.003 0.207 0.872 -0.131

Rather than attempting to interpret the individual figures, a histogram is

compiled to identify the variables that reoccur in the various annual models.

This histogram is shown in Figure 3.3.

(40)

Figure 3.3: Sales turnover most significant factor histogram

18 -i

It is evident from Figure 3.3 that the top 5 variables, measured in the number of occurrences in the top 5 for each year, are as follows:

• Inventory (16 times).

• Total asset turnover (11 times). • Operating profit margin (7 times).

• Capital requirements ratio (7 times); and • Return on equity (7 times).

Sales turnover growth

Table 3.8 contains a summary of the stepwise regression models for each year of the study for sales turnover growth, showing the adjusted coefficient of determination or adjusted R2 of each regression model. Considering the

depth of the stepwise regression for the various years, it is evident that the sales turnover growth is related mostly to only a few variables. The model fit

(41)

is generally significantly lower that that of sales turnover, as shown in Table 3.5.

Table 3.8: Sales turnover growth regression models

Year 11 Factor 10 Factor 9 Factor 8 Factor 7 Factor 6 Factor 5 Factor 4 Factor 3 Factor 2 Factor 1 Factor 1992 -.048 -.020 .007 .031 .054 .075 .093 .107 .103 .111 .100 1993 -.016 .012 .039 .063 .086 .108 .121 .133 .136 .148 .124 1994 .132 .154 .175 .193 .208 .204 .201 .188 .164 1995 -.152 -.123 -.096 -.071 -.052 -.033 -.021 -.030 -.026 -.011 .001 1996 -.121 -.093 -.066 -.041 -.018 .004 .022 .035 .045 .048 .023 1997 -.034 -.007 .012 .028 .046 .037 .039 .049 .045 .036 .024 1998 .868 .871 .871 1999 -.058 -.035 -.012 .003 .015 .005 .003 -.006 .001 .012 .004 2000 .391 .403 .414 .423 .431 .438 .442 .445 2001 .060 .078 .095 .105 .116 .120 .121 .130 .113 .094 2002 .418 .429 .438 .439 .438 .432 .428 2003 .089 .105 .119 .133 .144 .153 .149 .140 .135 .140 2004 .028 .045 .061 .076 .089 .100 .108 .111 .113 .112 .095 2005 -.093 -.075 -.059 -.043 -.027 -.013 -.001 .011 .013 .019 .023 2006 .319 .331 .341 .352 .362 .371 .367 .356 2007 .089 .102 .114 .125 .135 .143 .141 .130 .129

Refer to Table 3.9 for regression models using all eleven independent variables to explain sales turnover growth for 1992 to 2007, showing the standardized coefficients of each regression model. A multiple linear 11 factor regression model for each year results in a mean model fit of 11,7%, compared to a fit of 82.0% that was found for sales turnover. Thus, the models can generally only predict sales turnover growth to an accuracy of 11.7% on average. The most significant variable related to sales turnover growth is earnings per share, with a mean standardized coefficient of 0.239.

(42)

Table 3.9: Sales turnover growth 11 factor regression model Standardized Coefficients 1 _ CO CD > 3 K CD 3 .<£ c O ) 1— CO S o ha CL O ) c CO 1 -<D Q . O o CD tx CO C CD E CD i_ '5 a* a> DC 75 * ■ * 5. CO O LU o DC o "co DC C o c a> 4-1 a> DC c CO

s

o C L CD z < O DC > o c k. 3 * ■ * a> CO 10 CO "to o 1-CO CL LU CD i -CO .c CO L. CD Q . 3 o ts sz CO CO O L. O c CD > 2 CD > ■a c CD ■ a "> 5 1992 -,048 -.617 -.017 .209 .042 .190 .050 -.058 -.029 -.203 -.098 -.198 1993 -.016 .034 -.155 -.215 -.108 -.310 .065 -.031 .042 -.121 -.145 .052 1994 .132 .566 -1.660 .040 -.030 .393 -.376 .057 1.671 -1.111 -.167 -.322 1995 -0.15 .818 -.643 .802 .426 .013 -.512 .042 .157 -.177 -.016 .799 1996 -.121 -.433 .231 .000 .058 .109 -.167 .006 .085 -.137 -.189 ■ 01 1997 -.034 -1.452 .514 .189 -.092 1.103 .230 -.327 -.488 -.029 -.126 -.244 1998 .868 .530 -.301 -.224 .219 -.359 -.336 -.008 1.228 -.709 .292 -.048 1999 -.058 .082 -.330 .291 .115 -.079 -.166 -.205 -.447 .585 -.239 -.230 2000 .391 -.654 .108 .560 .048 .080 .725 -.056 -.327 .146 -.064 -.011 2001 .060 .136 -.289 -.140 .224 .017 -.038 -.098 .883 -.719 .311 .182 2002 .418 -.021 -.252 -.063 .269 .495 -.251 .147 .280 -.545 .209 -.524 2003 .089 .318 -.370 .325 -.025 .223 -.559 -.054 .196 -.348 -.013 -.056 2004 .028 -.211 -.135 .118 .078 .591 .041 -.008 -.026 -.018 -.101 -.068 2005 -.093 -.110 -.260 -.028 .037 .320 -.026 -.055 -.031 -.109 -.004 -.170 2006 .319 .251 .003 .375 -.099 -.521 981 .038 .279 -.103 .045 -.830 poo? .089 .983 -.603 -.035 -.021 -.266 -.483 .075 .343 -.410 .065 -.097 | [Mean 0.117 0.014 -0.260 0.138 0.071 0.125 -0.051 -0.033 0.239 -0.251 -0.015 -0.111

Investment guidelines based on future growth indicators