一籃子貨幣之資產配置與風險衡量

Abstract

台灣的人身保險業的保單多以長期為主，因此，在資產配置方面同時也需要較長期的投資工具來配合，就國內市場上而言，缺乏長期有效的投資工具且報酬率遠低於國外，造成國外投資的比重在保險業資金運用上逐漸攀升，但是，現階段保險業國外投資大多著重在美國，所以，本篇論文將針對保險公司把資金投資在更多不同國家，研究整體的投資組合是否比僅投資在台灣以及美國地區的績效要來得更佳。 此外，在風險衡量方面，首先採用傳統的數量方法以平均數─變異數為主，以變異數作為風險的衡量，隱含投資者對正、負報酬的關心程度是一樣，然而這並不符合實際情況，投資者真正關心的是下方風險的程度，所以，本文也採用VaR來衡量風險。 實證方面，本文利用十三個國家一籃子貨幣組合的方式來進行資產配置，發現透過國際資產間相關係數呈現中、低度相關，確實降低原本只投資在台灣與美國的資產風險，使整體投資組合之效率前緣更往外移，並在資產報酬率為常態分配的假設下，用標準差或是VaR當作衡量風險的指標所得到的效率前緣圖形並沒有太大的差異。

In Taiwan, life insurance policies issued are mainly long term; therefore, longer investment tools are required in asset allocation at the same time. For the domestic market, the shortage of effective long-term investment tools and the return of investment being far lower than foreign average made the percentage of foreign investment in respect of the available capital go up gradually. However, the insurance company lays more emphasis on the United States in terms of foreign investment. As a result, the thesis discusses whether the outcome would be better if the insurance companies put their money in more different countries instead of in Taiwan and the United States only. In addition, the traditional mathematical method—expected return-variance—is used to evaluate risk. The risk valuation based on variance implies that the expected returns, either positive or negative, are irrelevant to the investors, which does not conform to the present condition. In fact, the investors care about the degree of downside risk. Hence, the thesis uses VaR to evaluate the risk. In empirical, the thesis practices asset allocation by pooling a basket of currencies from 13 countries. By the analysis, the correlations between the international assets are moderate or low-correlated, indicating that the asset risk is reduced compared to the one invested in Taiwan and the United States only. The efficiency frontier of the portfolio moves outward. Under the assumption of normally distributed asset return, there is little difference in the efficiency frontier plot between the variance method and the VaR one used as risk evaluation indicators.