Median Response to Shocks: A Model for VaR Spillovers in East Asia

---

• 研討會日期 : 2015-12-01
• 時間 : 14:30
• 主講人 : Professor Giampiero M. Gallo
• 主持人 : Professor Yeutien Chou
• 地點 : Conference Room C103
• 演講者簡介 : Professor Gallo received his Ph.D. in Economics from University of Pennsylvania, Philadelphia in 1989. He is currently a Professor of Econometrics at University of Florence. His research interests are Financial Econometrics, Volatility Modeling, Multiplicative Error Models, Nonlinear Time Series, Forecasting and Model Validation.
• 演講摘要 : Value at Risk as a measure of a meaningful threshold in a conditional distribution of returns has gained relevance, especially in a risk management framework. More recently, attention was paid to the concept of Conditional Value at Risk as a measure of sensitivity of financial assets or markets to specific events in other assets or other markets. In this paper, we elaborate a procedure for analyzing interdependence across companies or markets assuming the presence of an entity which can be considered as originator of a shock of a certain size. We follow a route in which 　． we start from interpreting a given shock in the originator as corresponding to a certain level of volatility times the value of the quantile in the unconditional distribution of standardized returns as customary in the conditional variance literature; 　． we measure volatility as the daily range and its dynamics is modeled series by series (a GARCH can be used as a benchmark approach); 　． we then derive the features of several bivariate distributions between the originator and each impacted market from the standardized returns applying a copula function approach; 　． the estimated bivariate copula function (on the space (0; 1) x (0; 1)) will be shaped by the correlation between markets: for a given probability associated to a relevant event in the originator (VaR{defining threshold), we can read the median response on the vertical axis (relative to the impacted market). To get an idea about possible outcomes, think that such a conditional median will be smaller than one half, if the joint distribution has a high positive correlation; 　． such a probability is associated with its quantile in terms of standardized returns to derive the corresponding response from the shock; 　． the actual response is given by choosing what volatility prevails in the originator and calculating the corresponding average value in the impacted market. We apply our methodology to nine Asian markets (Malaysia, China, Taiwan, Hong Kong, Thailand, Indonesia, the Philippines, South Korea, Singapore) initially choosing Hong Kong as the originator of relevant shocks (as in previous papers of ours). The choice of originators can then change giving rise to different measures of the impact of a shock across a network of markets.