Information spillover across international real estate investment trusts: Evidence from an entropy-based network analysis

Qiang Ji, Hardik Marfatia, Rangan Gupta
2018 The North American journal of economics and finance  
In this study, we unveil information spillover between international real estate markets using an entropy-based network approach for real estate investment trusts (REITs). Our novel approach is simple and yet flexible enough to accommodate the nature and extent of information spillover among several components of the global housing network. For a network of nine leading industrial economies, we unveil static and time-varying information spillover of REIT returns using total transfer entropy,
more » ... rwise net transfer entropy and directional ("From", "To") transfer entropy. Evidence suggests that the greatest pairwise transfer entropy is from the US to Australia, whereas France, the Netherlands, New Zealand and Singapore are the largest information recipients in the network. The time-varying evolution of total transfer entropy also exhibits a declining trend, supporting the decoupling hypothesis for the global housing market network. The extreme value analysis shows the changing role of US and UK housing markets. JEL Codes: R30, R33, G14. REITs are also very good proxies for the real estate market, providing high-frequency observable data (Akinsomi et al., 2016) . Furthermore, while unsystematic risk arising out of the unique housing market dynamics can be minimised through diversification, the exposure to systematic risks depends on the degree of interdependency. Thus, it is necessary to estimate the information flows within the global REIT network both from the perspective of policymakers who try to navigate international currents and from that of international portfolio managers searching for diversification gains. In this study, we take advantage of the connectedness framework developed by Diebold and Yilmaz (2014) . In particular, we construct pairwise net transfer entropy within the network of nine industrial REIT markets, which include Australia, Canada, France, Hong Kong, the Netherlands, New Zealand, Singapore, the United Kingdom and the United States. 1 This provides us the weighted in-degree and out-degree information flow in the real estate network. We derive "From" and "To" transfer entropy to measure the information spillover among these network REIT markets. In order to take a holistic approach, a system-wide total transfer entropy is estimated. In our empirical modelling strategy, we acknowledge that the underlying structure of the global housing market has changed significantly over time, and we analyse time-varying information connections within the network. Using these apparatuses, we analyse the extent and nature of information spillover based on the extreme values of each country's REIT returns. The results show that the largest pairwise transfer entropy in REIT returns is from the US to Australia, but from Australia to the US, there is very little information flow. The Australian REIT market is the largest recipient of information, whereas Hong 1 These markets combined constitute 88.7% of the global REIT index, based on market capitalisation,
doi:10.1016/j.najef.2018.04.002 fatcat:jchg4i7xrjhoxguz25usr3c5oa