Multi-scale features of volatility spillover networks: A case study of China's energy stock market

The objective of this study is to examine the multi-scale feature of volatility spillover in the energy stock market systematically. To achieve this objective, a framework is proposed. First, the wavelet theory is used to divide the original data to subsequences to analyze the multi-scale features, and then the Generalized Autoregressive Conditional Heteroskedasticity model with Baba, Engle, Kraft, and Kroner specification (GARCH-BEKK) and the complex network theory are used to construct the spillover networks. Finally, the stock prices in the energy sector of China from 2014 to 2016 are used to conduct experiments. The main contribution of this paper is that we find various features of volatility spillover transmission in different time scales among energy stock prices. The results indicate that the volatility spillover effects are more fragmented in the short term, while the volatility changes will be only transmitted by a small number of important stock prices in the long term. In addition, we captured the key paths of volatility transmission by using the smallest directed tree of network under different timescales.