Browsing by Subject "Test of variance time function"

Now showing 1 - 1 of 1

Open Access
The Behavior of stock returns in Turkey: 1986-1988
(1989) Başçı, Erdem
This study investigates distributional and time series behavior of common stock returns in Istanbul Stock Exchange (ISE) for the period 1986-1988. The distributions of weekly price returns deviate from normality with sharp peaks, heavy tails and positive skewness. These observations are similar to those of United States stock markets but ISE returns have higher means and higher variances. The first order serial dependence is insignificant for most stocks and Box-Jenkins linear forecasting models shows a poor performance. So, published past price information cannot be used to obtain better forecasts of future prices by this model. This observation is in line with the random walk behavior as expected from a weak form efficient market. Applicability of Box-Jenkins models may be questioned however, since variance of returns is not stationary due to a second order dependence. This type of dependence is not against weak form efficiency and is seen in US stock returns as well. To detect any longer term dependence, the test of variance-time function is employed.:· The results indicate significant long term dependence for most stocks and this is against weak form efficiency. The weekly change in trading volume series turns out to be forecastable by univariate Box- Jenkins models and it seems to explain some of the variation in stock price returns.