The structure of REIT-beta

Abstract

Recent studies have documented an asymmetry in the market-beta of equity Real Estate Investment Trusts (REITs) based on high and low Gross Domestic Product (GDP) growth states, as well as in bull and bear stock markets. The asymmetry has been deemed a puzzle (Chatrath et al., 2000; Chiang et al., 2004); some previous studies explained it by describing the structural changes in REITs market and others included more variables to reduce the effect of asymmetry. What seems to be lacking, however, is a general theoretical explanation. This article provides a theoretical model in which the daily and monthly price series of REITs are separately described to explain the structure of REIT-beta and to solve this puzzle. We find there are four factors and the interaction of those determining the value of estimated beta. The results of previous studies might only be able to observe a few pieces of the nature of REIT-beta.

Keywords:

• asymmetry
• beta
• REITs
• frequency

JEL Classification::

• G11
• G20

Notes

¹ Those are factors like Gross Domestic Product (GDP), economic growth and interest rate. Payne (2003) also examined the influence of macroeconomic state variables on the excess returns of REITs. Payne identified the response of REIT excess returns to unexpected changes in the real output growth, inflation, term structure of interest rates and the federal funds rate using the generalized impulse response analysis.

² Those are factors like market size, and book-to-price (Fama and French, 1993).

³ λ is the coefficient reflecting the impact of trading on the stock price index.

⁴ Because those are all white noise items, it is not necessary to discuss those series.

⁵ Since the variable of economic growth data is quarterly data, we divide each quarter by three to get a monthly number.

⁶ The values of dependent variable are very small in our empirical analysis. Thus, we set time trend variable (t) in a way so that its value is consistent in scale with the value of dependent variable. The time trend variable is set to begin on 0.01 with an increment of 0.01 for each subsequent month. This setting is rather technical and the empirical result through this setting is unaffected.