本文研究不同投資組合策略的差異,有別於以往文獻大多使用債券、股票等資產,本研究以純REITs資產來組成投資組合以研究REITs資產近年來在資本市場的特性。由於金融資產具有厚尾高狹峰的特性,本文導入常態假設下的極小化變異數法、常態假設但考慮下方風險的極小化風險值法與只考慮下方風險的極小化共低偏動差法等三種以風險極小化為考量的投資組合策略做資產配置並檢視投資組合的投資績效。在靜態估計方面,發現極小化共低偏動差法的績效優於其他兩者,進一步改採動態滾動估計的方式,得到的結果與靜態估計一致,且績效較靜態估計的方法更好。另外根據相似性分析發現,在靜態估計中,極小化變異數法與極小化風險值法的差異不大,相似程度大部分都高達99%,極小化共低偏動差法與其他方法的差異則較大,且持有資產數通常也較其他兩種方法少。動態滾動估計也有類似的結果,但由於動態的估計期間相對較短,因此較能凸顯投資組合策略間較細部的差異。
Traditionally, most of literatures have discussed about the portfolio optimization, which focus on stocks or bond assets. Differ from these literatures, we use pure real estate securities to examine the best portfolio criterion. We use Min CLPM, Min variance and Min VaR portfolio criteria to optimal portfolio and compare their performance. We use eight largest and most developed REIT markets to examine the performance of portfolio strategies and the similarity between strategies. In static estimation, Min CLPM has better performance than other portfolio strategies. In order to ensure its robustness, we use dynamic rolling method to analysis portfolio criteria. The result is more robust and the performance has significant difference than static estimation. We further examine the similarity of the asset allocation. The analysis indicates that Min variance and Min VaR have similar allocation, most of them get 99% similar. On the other hand, the allocation of Min CLPM is different than other portfolio strategies and has less assets in the portfolio.