標題: 考慮景氣循環與風險值下的最適資產配置之實證研究
A Study of Optimal Asset Allocations with Business Cycle and Value at Risk
作者: 張吟綺
Yin-Chi Chang
許和鈞
Her-Jiun Sheu
經營管理研究所
關鍵字: 風險值;最適資產配置模型;景氣循環;GARCH;Value at Risk(VaR);Optimal Asset Portfolio;Business Cycle;GARCH
公開日期: 2004
摘要: Markowitz投資組合理論在資產配置模型早已被廣泛的運用,但在運用時卻有許多限制與假設不符合市場實際狀況,尤其是在實務上,財務資產的報酬分配呈現肥尾、群聚性與槓桿效果等現象。傳統投資理論使用變異數衡量風險,將資產價格正負波動均列入考慮,但投資人所關注僅為對其投資不利之下方風險,因此本研究採用考慮風險值的最適資產配置模型(Mean-VaR)進行實証研究。本研究以簡單平均法、指數加權平均法、GARCH模型與歷史模擬法等方法估算風險值,進而討論不同風險值估算法。下的配置績效是否有顯著差異。此外,由於資產價值會受到總體經濟變動影響,因此本研究也透過Mean-VaR模型,探討總體經濟變化對最適資產配置所決定的實際投資組合報酬有何影響。實證結果發現,景氣波動確實會影響報酬表現,在景氣擴張時期的報酬表現顯著優於景氣收縮時期,然而,四種風險值估算法無論在景氣好或景氣壞,透過Mean-VaR模型決定的最適配置比例所呈現的報酬表現並無顯著差異。最後,利用壓力測試與回溯測試檢驗風險值估算法的穩定性與可靠度,本研究也發現指數加權平均法的表現優於其他三種方法。
The portfolio theory of Markowitz is widely applied by financial practice nowaday. However, a lot of researchers found that there were many limitations and disadvantages in the model, including fat tails of asset returns, clustering of a asset volatility or leverage effects. Besides, traditional portfolio theory adopts variance to measure the risk of the market, i.e., it does not distinquish the difference between the values over/below mean value. For a lot of investors, downside risk affects most of their attentions. The concept of Value at Risk (VaR) is applied in this work. SMA, EWMA, GARCH and historical simulations are all employed in the caculations of VaR. The Mean-VaR framework is adopted to explore the asset alloactions. Since it is reported that the business cycle has powerfully influence on the movement of the asset prices, the Mean-VaR model is applied to study how the changes of macroeconomics affect the return performance. The results show that the business cycle has significantly influence on the performance of returns with more explanation power in the expansion stage than in the recession stage. However, it makes no significantly difference with any of the four methods in getting VaR.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT009237534
http://hdl.handle.net/11536/77288
顯示於類別:畢業論文