Skip navigation
Please use this identifier to cite or link to this item: http://arks.princeton.edu/ark:/88435/dsp01fq977t95w
Title: A COMPARISON OF MODELING TECHNIQUES FOR TIME-VARYING BETA IN ASIA-PACIFIC REAL ESTATE INVESTMENT TRUSTS
Authors: Wang, Lucia
Advisors: Fan, Jianqing
Department: Operations Research and Financial Engineering
Class Year: 2014
Abstract: Recent literature has suggested that real estate investment trust (REIT) betas, like stock betas, are time-varying in nature. However, despite the substantial literature confirming conditional REIT betas, there is a lack of research on how best to model them. Additionally, with the exception of the Australian market, Asia-Pacific REIT markets are relatively young, further contributing to the limited amount of literature. This thesis compares different modeling techniques according to in-sample estimates and out-of-sample forecasts for time-varying REIT beta in the Asia-Pacific region. The following models are evaluated: rolling regression, the multivariate dynamic conditional correlation (DCC) GARCH model, the Schwert and Seguin model, the Markov switching model, and the state space model with Kalman filter, and the following markets are included: Australia, Hong Kong, Japan, Malaysia, Singapore, and Taiwan. The results present the multivariate GARCH model as the best estimator for in-sample beta, but also find the Schwert and Seguin and Markov switching models to perform well. For out-of-sample forecasting, the results are consistent, finding the multivariate GARCH model as the best performer. However, in addition to the multivariate GARCH model, the rolling regression and state space model may also be well-suited for beta forecasting.
Extent: 68
URI: http://arks.princeton.edu/ark:/88435/dsp01fq977t95w
Type of Material: Princeton University Senior Theses
Language: en_US
Appears in Collections:Operations Research and Financial Engineering, 2000-2016

Files in This Item:
File SizeFormat 
Wang,Lucia, Final Thesis.pdf1.9 MBAdobe PDF    Request a copy


Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.