Bear market predictability in Singapore

Abstract

This paper investigates a variety of macroeconomic variables in Singapore and United States (US) that can be used to predict the bear market in Singapore. We use the parametric Markov-Switching model to classify the state of the market and conduct in-sample and out-of-sample tests to find out the useful macroeconomic variables. In addition, we propose a way to test the accuracy of our prediction as the testing of the accuracy of bear market predictions does not exist in prior research papers. We use the nonparametric Bry-Boschan algorithm and the moving average approach as the benchmark in calculating this accuracy. We find that TED spread and Singapore term spread are significant and have good predictive accuracy. A combination of these two variables in the multivariate model achieves on average an accuracy of about 67%. Our results also show that not all variables that are relevant and significant to the stock market index can predict the bear market accurately. Further, our results are in line with the extant literature that macroeconomic variables can predict the states of the stock market, generating satisfying performance.