Executive Summary. In this study, we offer a refinement to a return attribution method proposed by the pioneers of return attribution analysis. Returns for the aggregate portfolio are decomposed into selection and allocation contributions as originally presented. We introduce the use of a neutral effect, which aggregates to zero at the portfolio level, that insures proper interpretation of the decomposition of the sector returns of the portfolio into selection and allocation contributions. This refinement is particularly relevant to private real estate investment, where portfolio performance is measured against both an aggregate benchmark and benchmarks for sub-sectors. In addition, we suggest a methodology for performing multi-period attribution analysis. Further, we offer a new presentation format to report both single and multi-period return attributes.
The practice of evaluating manager performance has gained popularity in recent years with the development of techniques for decomposing returns into contributions from various types of manager activities.
Published descriptions of these techniques date back a little more than a decade. Brinson, Hood and Beebower (1986) pioneered this work, proposing a methodology that separates raw returns into a selection component and an allocation component. Brinson, Singer and Beebower (1991), Higgs and Goode (1993), Singer (1996b) and Dolan (1998) further elaborated this methodology.1 Recently, this methodology was applied to Canadian commercial real estate by Hamilton and Heinkel (1995). A variation of this methodology was publicized by Lieblich (1995) in The Handbook of Real Estate Portfolio Management.
These techniques decompose any excess return registered by a portfolio into two major categories of contributions by the manager. A manager shows superior property selection skills when holding individual property investments that outperform the market. Alternatively, the manager may hold average performing properties in particular sectors of the market, under-weighting a property sector when sector return is low and over-weighting it when return is high, to achieve superior performance. Under these circumstances, this manager is said to have sector allocation skills. A manager may have better than average skills in both selection and allocation, in either selection or allocation, or neither of the two.2
In this study, we offer a refinement to this return attribution approach in cases where portfolio performance is measured against both an aggregate benchmark and benchmarks for sub-sectors. We also suggest a modified presentation format to report both single and multi-period return attributes. This technique may improve the capacity of both managers and investors to assess and improve management skills and manager value added to the portfolio management process. Variations of this methodology are also discussed. Private commercial real estate returns are used to illustrate this approach.
Suppose a portfolio can be fully decomposed into n sectors (typically five sectors for commercial real estate: apartment, industrial, office, retail and others) for which performance benchmarks exist. Define the following:
Our attribution methodology calls for decomposing returns into the categories shown in Exhibit 1.3 Gross value added for the portfolio is simply the difference between the portfolio return (R) and the benchmark return (B). For an individual sector i, however, gross value added is the proportion of the portfolio return attributable to sector i (alRt) minus the proportion of the benchmark return attributable to sector i (biBi). This formulation follows the original approach by Brinson, Hood and Beebower (1986) and can be found throughout the literature. As displayed in Exhibit 1, the decomposition of excess return into attribution categories applies equally to each sector (or …