直接投資實體不動產之組合分析：來自不良資產之實證

A Portfolio Analysis of Direct Real Estate Investment: Evidence from the Non-Performing Assets

本文討論「直接投資實體不動產之投資組合效率」的兩個問題：(1)就規模來看，投資多少「數量或金額」的不動產，方可達到效率的投資組合？(2)就具有強烈異質性的不動產來說，「類型」或「區位」兩大特性，何者對於效率性的不動產投資組合形成有較大的助益？透過來自資產管理公司所提供不良資產中844筆不動產抵押品的橫斷面資料，以及投資組合模擬，Logistic迴歸模式的分析，主要發現：(1)就「數目規模」來說，投資組合的效果較個別不動產投資的效率性為佳，但任3個不動產組合的效果，並不顯著優於任2個不動產的投資組合；再者，就「金額規模」來說，組合價格與效率性呈現非線性的U型關係，亦即組合金額愈大或愈小時，都可形成效率的投資組合；(2)「區位」多角化之組合效率，優於不動產「類型」之多角化組合效率。上述的研究成果，預期對未來資產管理公司處分不良資產，或不動產投資信託基金之組合策略，將具相當之政策涵意。

In this paper, we discuss the portfolio efficiency of direct real estate investment. Unlike the concept applied to the financial portfolio theory, the price of real estate is informationally opaque and heterogeneity. The diversification of nonperforming asset (NPA) from Asset Management Company (AMC) is tested by the combination of size, property type and location. The main empirical results show: (1) The portfolio efficiency of two real estate objects domains other combinations. On the other hand, the relation of the portfolio efficiency and asset amount shapes a two-pole U curve. The optimal portfolio strategy of NPA can be achieved, even if the portfolio amounts are not large enough. (2) The more diversity of property region is, the more efficient the portfolio is. These findings have significant implications for the portfolio strategy of AMC’s NPA and Real Estate Investment Trust (REIT) in Taiwan. Diversification is a central principle of investing, and the real estate investment is initially viewed as a significant core of asset portfolios. Although Markowitz had published his seminal article on the modern portfolio theory (hereafter, MPT) in 1952, applying these concepts and results of MPT to the case of physical real estate investment must be revised. To an extend, every real estate asset is unique, and its trading market as compared with, say, large-capitalization stocks and bonds, is characterized by a relative lack of liquidity, large purchase size, and high transaction costs for properties that are fixed at some location and heterogeneous. However, these studies will not revisit the well-known arguments regarding the applicability of MPT to the physical real estate. Rather the goals of this article try to answer the following two questions. First, we consider two dimensions of “geographic location” and “property type” to the problems of diversification “within” portfolios of the directly real estate investments, and try to answer: Which diversification method of the “geography location” or “property type” is superior? Second, how much of the optimal amounts of the portfolios is good for the direct real estate investment? Our paper uses a very unique data sample, a mixed portfolio of non-performing assets (hereafter, NPA) in Taiwan, to test the portfolio efficiency of the direct real estate investment. After the Asian financial crisis of 1997, some authorities try to set up the asset management companies (hereafter, AMC) to solve the problems of NPA. AMCs usually take two methods, packaged-sell or securitization, to manage their NPA. Neither or any method is involved how to combine the efficient portfolio of real estate. The empirical results of our study directly benefit on how managers of AMCs might reconsider their business strategy of NPA portfolio. An extension contribution is to find the evidence of small size, for example, only individual or two objects, in most REIT cases of Taiwan. We have two main findings. First, the diversification of location can form a more efficient portfolio in our Non-performing assets case of Taiwan. Second, the simulations of portfolios show that bigger is not always better. We find the relation of the portfolio efficiency and asset size shapes a two-pole U curve. The optimal portfolio strategy can be achieved, even if the portfolio amounts are not large enough.