Bayesian REIT volatility estimation and institutional portfolio allocation

Volatility estimation is an integral part of institutional finance with applications in risk management and portfolio allocation. Real estate investment trust volatility is examined using a Bayesian asymmetric GARCH model and is found to better estimate the true volatility series than does the traditional maximum likelihood approach. This paper discusses the shortfalls of maximum likelihood (ML) estimation and the advantages of the Bayesian estimation, particularly to real estate. Conditional variance estimation uncertainty is found to increase with volatility. A portfolio allocation problem highlights that the Bayesian approach performed better than the ML method in preserving capital.