Abstract
The purpose of this study is to establish the determinants of stock market returns for the UK market and to forecast the real capital gain for the Financial Times All-share index using monthly data for the period 1980–1994. Three models are used: (a) a model based on Fama's approach including expectations' variables for the growth of GDP and inflation; (b) a model where short-run and long-run impacts are separately treated and (c) an ARCH model where volatility in returns is modelled using conditional variance. One-period ahead forecasts from the general autoregressive-distributed lag model and ARCH model are compared with actuals for the post-sample period (December 1992 to November 1993) and we find that the latter model has a higher predictive power than the former.