Property Portfolio AnalysisGive the data requirements of the Markowitz model, it is practically applicable mostly at the metropolitan area and property type levels. This is feasible because historical indices of real estate investment performance for the 50 largest metropolitan areas and four major property types (office, retail, industrial and apartments are available from various vendors specialized in collecting and processing such data. Thus, the investment universe for the application of the Markowitz model in real estate may include many opportunities specified in metropolitan area/property type dimensions (for example, Dallas office, Houston office, Atlanta Retail, etc.).
Early applications of the Markowitz model in property portfolio analysis focused on portfolios including the four major property types at the national level. The return and risk measures, as well as pairwise correlation coefficients, were derived using historical return data taken from the database of the National Council of Real Estate Investment Fiduciaries (NCREIF). Such models verified that the argument that diversification of real estate portfolios across property types does provide diversification benefits that help reduce portfolio risk. Other applications have focused on the diversification benefits across locations using measures of economic performance. It should be noted that real estate return and risk measures derived through typical statistical procedures from historical data may be problematic when using the Markowitz model, because it is assumed that their underlying distribution is normal. In other words the Markowitz model mathematics assume that the return measures that enter the model are normally distributed around their mean. According to empirical work (Young and Graff, 1995), however, real estate returns are not normally distributed around their mean.
|
