This study investigates the ability of a broad range of 'style-based' characteristics to forecast monthly returns on JSE Securities Exchange SA shares. After being pruned for thinly traded stocks, the complete McGregor/Bureau of Financial Analysis (McG/BFA) JSE database is analysed using monthly cross-sectional regressions over the decade of the 1990s. A number of interrelated 'value' effects (price-to-NAV, dividend yield, price-to-earnings, cash flow-to-price and price-to-profit) together with a small firm effect are documented. All permutations of individually significant characteristics are then tested in a multifactor setting. The analysis suggests a two-factor (size and price-to-earnings) style-based model of expected returns on the JSE.
Directors are generally assumed to be better informed than the average outside investor about their company's value and prospects. Therefore when directors buy and sell the shares of their firm, they might be transmitting useful trading signals. This paper investigates whether directors of JSE Securities Exchange SA listed companies earn abnormal returns from own-company trades. If so, outside investors may share in this outperformance through mimicking these trades. An event study shows directors do earn substantial abnormal returns, but only from sale transactions, not purchases. However, most of this outperformance is due to the existence of market effects and not because of the director's transaction.
This study examines the reliability of earnings forecasts that companies provide prior to listing 506 companies that listed on the JSE Securities Exchange SA from the 1st January 1980 to the 31st December 1998 are analysed. Earnings forecasts reflected in the company prospectuses were compared with actual corresponding earnings. The study shows that pre-listing earnings forecasts were inherently understated by approximately 14, 3% and that company size, JSE sector and GDP growth had significant associations with the forecast error.
We investigate and compare the application of the Merton model (1974) and the Shimko, Tejima, van Deventer model (1993) to the valuation of risky debt in the South African context. In the Merton model a flat term structure for the short-term interest rate is assumed. We then consider the effect of stochastic interest rates, specifically the Vasicek model, as applied to risky debt evaluation by Shimko et al. We use the credit spread as a basis for testing the two models against empirical data. Much improved credit spreads are obtained by the Shimko <I>et al.</I> model.