An investigation into the extent to which a share price is influenced by earnings per share, dividends per share and cash flow per share of industrial companies listed on the Johannesburg Stock Exchange

Mara-Tjingaete, Cashandra Candice (2001-12)

Thesis (MBA)--Stellenbosch University, 2001.

Thesis

ENGLISH ABSTRACT: This research involves the study of the relationship between share prices, earnings per share, dividends per share and cash flow per share. The aim is to estimate the extent of the effect of earnings per share, dividends per share and cash flow per share on share prices. A sample of industrial companies, all listed on the Johannesburg Stock Exchange is used. The sample includes 17 sectors, containing a total of 326 companies. The data relating to these companies is sourced from the database of the University of Stellenbosch Business School. Twenty consecutive years of reported financial results are used. This includes (i) share prices as at the company's year-end, (ii) earnings per share, (iii) dividends per share and (iv) cash flow per share for the financial year. The data for each individual company is arranged in three separate structures. One structure, known as levels, is created using the data in their original form. A second structure, first differences, is constructed using the annual changes in the original data. The third structure, first difference percentages, is created using the annual changes in the original data calculated as a percentage of the previous year's figure. Several methods are used to analyse data, by way of statistical models known as linear regression, multiple regression and stepwise regression, which is a variation of multiple regression. Regression analysis is used in deterministic models because the value of one variable is predicted on the basis of other variables. Such models are called deterministic because they allow users of data to determine the value of the dependent variable from the value of the independent variable(s). Linear regression is capable of calculating the relationship between variables, which are dependent on each other and represents the dependency as a percentage. Multiple regression is a statistical model, which calculates and represents the relationship between three or more variables as an equation. Multiple regression is capable of calculating the relationship between the three or more variables which are dependent on each other and represents the dependency as a percentage. The statistical models are applied to each company individually, based on the three structures of the data. The conclusions of this research are based upon the results of the dependency of the variables, calculated by the various statistical models. The conclusions arrived at were that (1) the first differences structure is best suited to this model, (2) by placing the data of individual companies together in one pool, the problem of weighing each company was overcome, and (3) using linear regression results together with the multiple and stepwise regression results added value in terms of comparison of the accuracy of results. A comparison was drawn between the findings of this study and those of previous writers and is referred to in the text. Four researchers namely Strong, Board and Day, Brown and Mara-Tjingaete, regressed the results of earnings versus share price and produced R2 values of 12,8 percent, 13 percent, 7 percent and 7,8 percent respectively. Brown and Mara-Tjingaete found R2 results of 14 percent and 0,6 percent respectively for dividends versus share price and this study also produced an R2 result of 0,4 percent for cash flow versus share price.

AFRIKAANSE OPSOMMING: Hierdie navorsing behels die studie van die verwantskap tussen aandeelpryse, verdienste per aandeel, dividende per aandeel en kontantvloei per aandeel. Die doel daarvan is om die uitwerking van verdienste per aandeel, dividend per aandeel en kontantvloei per aandeel op aandeelpryse te bereken. 'n Steekproef van industriële maatskappye genoteer op die Johannesburgse Effektebeurs is gebruik. Die steekproef strek oor 17 sektore en sluit 326 maatskappye in. Die data van hierdie maatskappye is verkry van die databank van die Universiteit van Stellenbosch Bestuurskool. Twintig opeenvolgende jare se gepubliseerde finansiële state word in die steekproef gebruik. Dit sluit in (i) aandeelpryse soos op die maatskappye se finansiële jaareinde, (ii) verdienste per aandeel, (iii) dividende per aandeel en (iv) kontantvloei per aandeel vir die finansiële jaar. Die data vir elke maatskappy is in drie verskillende strukture gerangskik. Een struktuur, bekend as vlakke, is geskep deur die data in hulle oorspronklike vorm te gebruik. 'n Tweede struktuur, eerste verskille, is saamgestel deur die jaarlikse verandering in die oorspronklike data te gebruik. Die derde struktuur, eerste verskille-persentasies, is geskep deur die jaarlikse veranderinge in die oorspronklike data, bereken as 'n persentasie van die vorige jaar se syfer, te gebruik. Verskeie metodes word gebruik om data te ontleed, deur middel van statistiese modelle bekend as lineêre regressie, meervoudige regressie en stapsgewyse regressie, 'n variasie van meervoudige regressie. Regressie ontledings word gebruik in bepalende modelle omdat die waarde van een veranderlike gebruik word as basis om die waarde van die ander veranderlike te probeer voorspel. Sulke modelle word bepalend genoem vanweë die feit dat dit gebruikers van data in staat stel om die waarde van die afhanklike veranderlike te bepaal deur middel van die waarde van die onafhanklike veranderlike(s). Lineêre regressie kan die verwantskap tussen interafhanklike veranderlikes bereken en die afhanklikheid as 'n persentasie uitdruk. Meervoudige regressie is 'n statistiese model wat die verwantskap tussen drie of meer veranderlikes as 'n vergelyking bereken en voorstel. Meervoudige regressie kan die verwantskap tussen drie of meer interafhanklike veranderlikes bereken en die afhanklikheid as 'n persentasie uitdruk. Die statistiese modelle word toegepas op elke individuele maatskappy, gebaseer op die bogenoemde drie strukture. Die bevindinge van die navorsing is gebaseer op die resultate van die afhanklikheid van die veranderlikes soos deur die verskillende statistiese modelle bereken. Die uiteindelike gevolgtrekking was dat (1) die eerste verskille-struktuur die mees gepaste vir hierdie model is, (2) deur die data van die afsonderlike maatskappye in een poel te plaas, is die probleem van die gewig per maatskappy oorkom, (3) die gebruik van lineêre, meervoudige en stapsgewyse regressie het baie waarde ten opsigte van die vergelyking van akkuraatheid van die resultate toegevoeg. Vergelykings is getref tussen bevindinge in hierdie werkstuk en die van ander navorsers. Vier navorsers, naamlik Strong, Board en Day, Brown en Mara- Tjingaete het R2 resultate vir verdienste per aandeel teenoor aandeelpryse van onderskeidelik 12,8 persent, 13 persent, 7 persent en 7,8 persent gekry. Brown en Mara-Tjingaete het R2 resultate van onderskeidelik 14 persent en 0,6 persent bevind vir dividende per aandeel teenoor aandeelpryse en hierdie studie het ook R2 resultate van 0,4 persent vir kontantvloei per aandeel teenoor aandeelpryse gelewer.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/52114
This item appears in the following collections: