Modellering van die groei in jaarlikse verdienstesyfers van genoteerde Suid-Afrikaanse nywerheidsmaatskappye : 1974 tot 1993

Botha, Lomeus Jacobus (1995-12)

Thesis (MBA)-- Stellenbosch University, 1995.

Thesis

ENGLISH ABSTRACT: The price of shares is determined primarily by investors' current expectations about the future values of variables that measure the relevant aspects of a company's performance and profitability, particularly the anticipated growth rate of earnings per share. Empirically, no model estimated with only historical senes data has been found to have greater forecast accuracy than the random walk model in estimating earnings one period ahead. This has led to the conclusion that past and future earnings growth is uncorrelated and that only year t-l earnings are useful in forecasting year t earnings. Research by Mozes in the USA has found the opposite and his model is applied to the South African situation. The aim is to determine whether the Mozes model has greater forecasting accuracy in the prediction of earnings per share than the random walk model. The present study shows that the Mozes model has greater forecast accuracy in the prediction of earnings per share than the random walk model if the following criteria are met: the company must be classified as a large company in terms of market capitalisation; or the percentage increase in earnings per share must be large; and the earnings per share must be classified in the growth mode. It is demonstrated that if these criteria are met, the historical growth in earnings and the future growth in earnings are positively correlated and not distributed at random. If earnings per share is classified in the non~growth mode, the random walk model is more accurate in the prediction of earnings per share than the Mozes model and as such, only the earnings per share of year t-l is important in forecasting year t's earnings per share. The most important conclusion from the study is that earnings per share in the South African market is not always randomly distributed.

AFRIKAANSE OPSOMMING: Die prys van aandele word primer bepaal deur beleggers se huidige verwagtinge rakende die toekomstige waarde van veranderlikes wat relevante aspekte van die maatskappy se prestasie en winsgewendheid beinvloed, meer spesifiek die geantisipeerde groei in verdienste per aandeel. Empiriese studies het bevind dat die toevalslopie-model die grootste akkuraatheid in die vooruitskatting van verdienste vir een periode in die toekoms lewer indien van historiese tydreeksdata gebruik gemaak word. Die gevolgtrekking word dus gemaak dat groei in verdienste van die verlede en die toekoms nie gekorreleerd is nie en dat slegs jaar t-1 se verdienste belangrik is in die vooruitskatting van jaar t se verdienste. Navorsing deur Mozes in die VSA het die teendeel getoon en die model is in die ondersoek toegepas op Suid-Afrikaanse data om te bepaal of dieselfde bevindinge geld. Resultate van hierdie studie toon dat daar aan die volgende kriteria voldoen moet word alvorens die Mozes-model meer akkurate vooruitskattings van verdienste per aandeel lewer as die toevals-Iopiemodel : -die maatskappy behoort as 'n groot maatskappy geklassifiseer te wees volgens markkapitalisasie; of -die persentasieverandering in verdienste per aandeel behoort groot te wees; en -indien verdienste per aandeel as synde in die groeifase geklassifiseer is. Indien aan die kriteria voldoen word, is aangetoon dat historiese groei in verdienste en toekomstige groei in verdienste gekorreleerd is en nie ewekansig versprei is nie. In die gevalle waar verdienste per aandeel as synde in die nie-groeifase geklassifiseer is, lewer die toevalslopie-model oorheersend meer akkurate vooruitskattings van verdienste per aandeel as die Mozes-model en gevolglik is daar bevind dat slegs jaar t ~ 1 se verdienste per aandeel belangrik is vir die vooruitskatting van jaar t se verdienste per aandeel. Die belangrikste afleiding vanuit die studie is gevolglik dat verdienste per aandeel in die SuidAfrikaanse mark nie in aile gevalle sonder meer ewekansig versprei is nie.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/54792
This item appears in the following collections: