MULTI — SECTORAL TEMPORAL MODELS FOR GHANA STOCK RETURNS

Abstract

Stochastic and econometric models have earlier been used in studying or modeling movements of stock prices and returns' depending on the interest of the researcher. But econometric models such as the capital asset pricing model have been found useful in explaining the determinants of returns on investments in stocks. Cross-sectional analyses were performed to identify the economic variables that contribute significantly to the expected returns of individual equities listed on the Ghana Stock Exchange. The least-squared second differencing approach of disaggregation gave the best results in transforming the data into quarterly form. Although the central limit theorem in probability theory justifies the approximation of large sample statistics to the normal distribution in controlled experiments, using the specified distribution of the residuals eliminate biased regression coefficients and Cauchy distribution gave the best fit for the residuals of stock returns. Finally, a longitudinal model with Banded Toeplitz variance-covariance structure which incorporated Cauchy as the residual distribution was proposed for the Ghana stock returns.