Partial fulfillment for award of Master of Commerce

-

-

Time series of financial and economic data have been found to exhibit unique statistical properties not found in other time series. The statistical properties of the NSE 20-share index returns are examined over a period of 16 years, from January 2 1992 to December 24, 2007. In addition, I use the General Autoregressive Conditional Heteroskedasticity (GARSCH) model of Boollerslev (1986) to estimate the volatility present in the NSE-20 index returns. This period is marked by significant growth in market capitalization and structural change, with increased usage of information technology and increased regulation of the financial and money markets. I find that 1. the index returns display evidence of predictability, but the degree of predictability has reduced over time. 2. The index returns are non-normal, with significant skewness and leptokurtosis; 3) there is evidence of volatility clustering in the returns, where the periods of high volatility are followed with periods of low volatility and periods of low volatility are followed by periods of high volatility. 4. The frequency of fluctuations or volatility has increased over time. the statistical and volatility properties of returns from the market have evolved as the Nairobi Stocks Exchange has developed over time.