The Effect Of Integration On Investment Portfolio Diversification In East African Stock Exchanges

ABSTRACT The purpose of this study was to analyze the effect of integration on stock market investment portfolio diversification in East African stock exchanges. The study was based on the Nairobi Securities Exchange, Dar-es-salaam Stock Exchange and Uganda Securities Exchange with the target population of all the companies whose common stocks were traded over the sample time period. Covering a period of 100 months between January 2000 and April 2008, the researcher analyzed the pair-wise index correlation structure, co-movement and cointegration of the markets‟ Paasches index series. From the daily price lists, month-end trade volumes and prices of each market‟s trading stocks were extracted, forming the input for the index construction formula. The next step was to test each index for stationarity using the unit root test. The indexes were then paired and divided into pre-and post-breakpoint components with pair-wise deletion of the cases not completely missing at random, using the Customs Union Date (February 29, 2004) as the breakpoint. Accordingly, the sub-period data was used for the analysis of pair-wise comovement structure and market contagion after the application of the breakpoint event, the sign-up of the East African customs union treaty. The final stage of the data analysis involved estimating least squares regression lines. From each market pair, one index series was regressed on the other and the results used for the computation of the regression error and white noise with a one-month lag length. The researcher then tested the regression and residual error variances for unit root stationarity. The findings asserted both co-movement and divergence of the index pairs in the first and second study sub-periods, signifying irregularity of the investment diversification benefits in both cases. Further, each index pair was found to have reacted in its own way to the economic shock distorting the dependency structure. Finally, all least squares regression and residual error variances were non-zero proving that there was no pair-wise cointegration and hence no evidence of diversification debenefit in the long run, in the three securities markets.