Exploring capitalization rate differentials across property types

Real Estate Issues, Dec 1997 by Sivitanides, Petros S, Sivitanidou, Rena C

The understanding of how asset market behavior differs across property types is important for

institutional investors contemplating property-type diversification strategies.

INTRODUCTION

This article presents the results of analytical work intended to empirically identify differences in transaction-based capitalization rates across office, warehouse, retail, and apartment properties during the period of 1986-1996. Three types of differences in capitalization rates across these property types are investigated: first, differences in their fixed (timeinvariant) component; second, differences in the persistence of their time trends or the speed by which they adjust in response to changes in market conditions; and third, differences in the pattern of their intertemporal variations.

The understanding of how asset market behavior differs across property types is important for institutional investors contemplating property-type diversification strategies. An intelligent formulation of such strategies requires assessment of the differential return prospects of each property type. Such return prospects are determined in both the space (tenant) market, in which the time path of vacancies and rents is shaped, and the asset market, in which property prices are set. Capitalization rates are important determinants of the latter. A better understanding, therefore, of how they differ across property types can help investors better assess differential return prospects across property types. Although existing empirical studies have detected fixed differences in capitalization rates across property types, they have neither accounted for differential persistence nor examined differences, if any, in time trends.1 Examining aspects of such differential asset market behavior in an integrated fashion will set the platform for more accurate estimates of the different effects.

The second section of this article focuses on the empirical methodology employed in exploring the issue at hand. The third section elaborates on the analysis results and advances potential explanations for the sources of the empirically identifiable differences in capitalization rates across property types. Finally, the fourth section summarizes the conclusions of the article and discusses potential avenues for future research.

THE EMPIRICAL FRAMEWORK

Recent metro-specific data from the National Real Estate Index (NREI) point to non-trivial cross-section and temporal differences in transaction-based capitalization rates across four property types: retail, office, warehouse, and apartments. A cursory examination of capitalization rate patterns across these property types is insufficient in evaluating their statistical significance and magnitude. Thus, a simple empirical model, similar in spirit to models used to examine the differential behavior of vacancy rates, price appreciation, and real estate returns, has been formulated to help validate the statistical significance of the observed differentials.2 Following the aforementioned modeling framework, the capitalization rate for a given property type at any point in time t can be decomposed into a fixed property-type specific component, aj, and a random fluctuation around this component, t:

Equation 1

C^sub it^ = a^sub i^ E^sub it^

The fixed component represents that component of return that compensates the marginal investor for each property type's idiosyncratic risk characteristics. The random term, also allowed to vary across property types, reflects deviations from this fixed component due to market-based income growth expectations, as well as additional market-driven risk premia. Random market movements generate time variations in such income growth expectations and risk premia, thereby influencing the capitalization rate required by investors. For given rents, such new capitalization rates are established through adjustments in asset prices. Such asset price adjustments, however, may be hampered by several asset market inefficiencies. The latter include high transaction and adjustment costs; lengthy institutional decision-making processes that may prevent investor entry/exit; and informational inefficiencies hampering the buyer-seller matching process, especially in heterogeneous asset markets. It may thus take more than one period before transaction-based capitalization rates fully reflect the effect of random market movements. As a result, a fraction, p, of each period's random deviation from ai may persist into the next. The random component of the capitalization rate, Eit, can thus be expressed as in Equation 2, where both, Fi (t), denoting the random time trend, and pi, obeying O=

Equation 2

Et = F^sub i f(t) p^sub i^ E^sub it-1^ V^sub it^ Combining Equations 1 and 2 yields the empirical formulation in Equation 3 which sets the appropriate platform for analyzing potential differences in the behavior of capitalization rates across property types.

Equation 3

C^sub it^ = a^sub i^ F^sub i^(t) P^sub i^E^sub it^ v^sub it^