International stock market contagion: A CEEMDAN wavelet analysis

Abstract

This paper investigates the contagion effect among stock markets (Asia, European and America) under time varying frequencies by use of a CEEMDAN wavelet (complete ensemble empirical mode decomposition with adaptive noise) model. Firstly, we decompose stock index return into different independent intrinsic mode functions and wavelet decomposition functions. Secondly, we reconstruct the independent IMFs and wavelet decomposition functions into three components: a high-frequency component (effects of irregular events), a low-frequency component (effects of extreme events) and the long-term trend. Thirdly, we assess the accumulated impulse response and analyze the stock markets contagion under time- varying frequencies. Results show that shocks caused by irregular events and extreme events can be transmitted between different stock markets. Moreover, shocks caused by irregular events can pose sudden and short-term risk contagion to stock returns. And shocks caused by extreme events can pose positive and sustained risk contagion to stock returns. Also, we compare our results with those from the discrete wavelet transform model and findings in the literature.