An approximate entropy approach to examine the non-linear dependence in daily Indian exchange rates

The purpose of the study is to examine whether the returns and volatility for Indian exchange rates possess non-linear dependence. Furthermore, an attempt is made through a rolling-window approach to check whether non-linear dependence is time-varying. The study employs approximate entropy statistics to examine the non-linear dependence. We also estimate the Tsay statistics to test for non-linearity. The empirical results provide the evidence of strong non-linear dependence in the Indian exchange rate returns and volatility and also that is time-varying. The results also suggest that the GARCH model, which has been used in the study, is misspecified. The evidence of non-linearity has serious implications for asset pricing, risk management and policy making.