不同風險值模型在亞洲金融市場之應用

Abstract

雖然風險值在金融界已被廣泛的採用為風險管理的工具，但其中有些假設卻在這些年來廣受批評，而條件風險值的提出，解決了傳統風險值無法考慮資產報酬厚尾特性的缺點。本研究計算風險值及條件風險值兩種指標，採用了二種常用的風險值衡量方法– 變異數共變異數法及歷史模擬法，藉由利用亞洲十個國家的股票市場歷史資料，將各股票市場依據風險值的大小作排序，本研究有下列發現：（1）採用變異數共變異數法與歷史模擬法產生不同的排序結果，而排序的差異幅度與市場的波動度有正向的關係；（2）當信賴水準越低，不同風險值模型產生的排序結果差異越大；（3）在市場波動度大的期間，風險值與條件風險值產生的排序差異越大。

Although Value at Risk (VaR) has been broadly welcomed by financial market participants as a risk management tool, some of its assumptions are widely criticized in the academic literatures. Conditional-VaR (CVaR) is a risk measure intended to capture the fat-tail phenomenon of the distribution of asset returns which is not incorporated by traditional VaR measures. This study adopts two widely used approaches, the parametric approach and the historical simulation approach, to calculate VaR and CVaR measures and make comparisons among these measures and approaches. Using historical data for ten Asian equity markets, this study found that (1) the historical simulation approach has different risk ranking results from parametric approach, and the degree to difference is related to price stability; (2) when lower confidence level is selected, different VaR models yielded a greater difference in risk ranking results among countries, (3) VaR and CVaR give different ranking order results especially during periods of financial turmoil.