Modeling of Returns and Trading Volume by Regime Switching Copulas

Henryk Gurgul, Artur Machno, Roland Mestel
2013 Managerial Economics  
The dependence between the stock markets can be measured through such variables as stock return, trading volume and volatility. The most frequently used methodology in the investigations of the interdependencies is based on Granger causality and VAR model (see [18] ). In one of the earliest contributions on dependency of stock markets Eun and Shim ([15]) checked the relationships among nine major stock markets including Australia, the UK and the US by means of the Vector Autoregressive (VAR)
more » ... el. The authors found that news in the US market has a major impact on the other markets. Lin et al. ([31]) focused on interdependence between the returns and volatility of Japan and the US market indices using high frequency data of daytime and overnight returns. They established that daytime returns in the US or Japan market were linked with the overnight returns in the other. Kim and Rogers ([27] ) studied the dynamic interdependence between the stock markets of Korea, Japan, and the US. They underlined the importance of Japanese and the US stock markets for Korean market since the last became more open for foreign investors. By mean of EGARCH model Booth et al. ([11]) found strong interdependence among the Danish, Finnish, Norwegian and Swedish Stock Market. According to the authors the essential dependence started with the so-called Thailand currency crisis. However, it was not observed after the Hong Kong crisis. Ng ([33]) established significant causality running from the US and Japan stock market to six Asian markets: Hong Kong, Korea, Malaysia, Singapore, Taiwan and Thailand. Klein et al. ([28]) by means of wavelets technique, applied to three developed markets: US, Germany and Japan and two emerging markets Egypt and Turkey proved that changes in these developed markets had effects on the emerging markets. In the paper [6], using the VAR-EGARCH model, it is checked the interdependence among three EU markets namely Germany, France and the UK. The results supported the hypothesis of the cointegration among the mentioned stock markets. Sharkasi et al. ([42]) used wavelet analysis and found the global co-movements among seven stock markets, three in Europe (Irish, UK, and Portuguese), two in the Americas (namely US, and Brazilian) and two in Asia (Japanese and Hong Kong). The contributions by Ammermann and Patterson ([2]), Lim et al. (30), Lim and Hinich ([29]), Bessler et al. ([7]) or Bonilla et al. ([10]) tried to established a different pattern of the stock price development. The authors detected long random walk phases. They alternated with short ones and showed significant linear and/or nonlinear correlations. The contributors thought that these serial dependencies had an episodic character. Due to these contributors the serial dependencies caused the low performance of the forecasting models. Nivet ([34])
doi:10.7494/manage.2013.13.1.45 fatcat:i2luxctporbavoamrvc5yd3uim