Empirical Investigation of the Nature of Returns of Stock Prices of three Prominent South Asian Markets using Parametric and Non Parametric Techniques

Abstract

The present study makes an attempt to test whether the returns of the three major indices of the prominent South Asian Markets namely BSE Sensex, CSE ASPI and Pakistan KSE 100 follow aStationary Process.The month wise closing data has been collected for the above indices and period of the study is eleven years, April 1, 2005 –March 31, 2016, The data has been log transformed to first difference (ln.Pt - ln. Pt-1 )and all the tests have been applied on log transformed data.Both Parametric and Non Parametric tests have been employed for testing the hypothesis of Stationarity. The parametric tests include Augmented Dickey Fuller test which has been traditionally used for checking the non-random character of time series, Dickey Fuller Generalized Least Squares (DF GLS of Elliott, Rothenberg, and Stock (1996) test,Box Pierce(1970) ‘Q’ statistics, & Variance Ratio technique of Lo and Mac Kinlay(1988). The nonparametric tests include turning point test, the difference of the runs test & KPSS(1992)test. The hypothesis has also been tested graphically using autocorrelation and partial autocorrelation techniques The variance ratio tests for randomness is applied first by assuming homoscedasticity or constant variance, & later by making it robust after incorporating heteroscedasticity in time series. The results of our study as revealed by parametric tests (ADF, ‘Q’ Statistics & Lo and Mac Kinlay Variance Ratio tests) confirm that the returns of, CSE ASPI is stationary, KSE is found to be stationary in three out of four tests, while BSE Sensex is stationary in only two of the four tests.. Coming to non-parametric results, runs test and KPSS test support stationarity of returns of all our indices. The present study shows that testing of stock returns for stationarity using only one single test is not at all conclusive &a good research must combine two-three parametric and one-two non parametric tests to get the satisfactory result w.r.t. stationarity of a variable.

Keywords: Variance Ratio Test, Random Walk, Homoscedasticity, DF GLS test, South Asian Markets

View PDF