Identifying Regime Shifts in South African Exchange rates

Linear time series models are not able to capture the behaviour of many financial time series, as in the cases of inflation rates, exchange rates and stock prices data. To overcome this problem, nonlinear time series models are typically designed to capture these nonlinear features in the data. In this paper, we use portmanteau test and likelihood ratio test (LR) test to detect nonlinear feature and to justify the use of 2-regime Markov switching autoregressive model (MS-AR) in South Africa exchange rate between 1995 and 2013. For model selection criteria (AIC and SBC) were used and for identifying best model error matrix such as MEA and MSE were used. The study compared the in-sample fitting between linear model and Markov switching model. From the error matrix (MEA and MSE) values, it is found that the MS ?AR(3) model is the best fitted model for exchange rate. In addition, the regime switching model also found to perform better than simple autoregressive model in in-sample fitting. This result justified that nonlinear model give better in-sample fitting than linear model.