Recursive modelling in predicting excess returns - case of the Nairobi Securities Exchange

Abstract

The aim of this study was to test the applicability of a recursive modelling approach in modelling stock market returns in the Nairobi Securities Exchange. The dependent variable was the Nairobi Securities Exchange All Share Index, with a core assumption being that firms do not pay dividends. I test the applicability of recursive modelling using three returns models, each containing different regressors, and compare the performance of the models in predicting future values of the index, as well as the performance of the recursive forecasting model compared to a dynamic forecasting model. I find that a recursive model is capable of predicting future values of the index using all three models, with varying performance among the models, but fail to find conclusive evidence to suggest that the recursive forecasting model significantly outperforms a dynamic forecasting model.