Can the Sharia-based Islamic stock market returns be forecasted using large number of predictors and models?

Abstract

This study employs 14 global economic and financial variables to predict the return of the Islamic stock market as identified by the Dow Jones Islamic Stock Market (DJIM). It implements alternative forecasting methods and allows for nonlinearity in the multivariate predictive regressions by estimating time-varying parameter models. All the methods fail to forecast the returns of the Sharia-based DJIM index over the out-of-sample period. The forecasts are weak at best, with only four predictors, the 3-month Treasury bill rate, inflation, oil price and return on the S&P500 Index, outperforming the benchmark autoregressive model of order one. The study suggests that the DJIM return is best predicted by an autocorrelation(1) model, and that future research should aim at analysing whether the performance of the linear autoregressive model can be improved by using nonlinear methods.