Estimating value at risk in Saudi real estate investment trusts: A garch-based approach

Abstract

This paper compares the performance of the GARCH (1,1) and GJR-GARCH (1,1) models in forecasting VaR for Saudi REITs across three distinct periods: the pre-COVID period (2016–2019), the during-COVID period (2020–2021), and the post-COVID period (2022–2024). The study estimates log returns and models them using GARCH-type structures, applying the Kupiec test for backtesting the VaR forecasts. The results show that both GARCH (1,1) and GJR-GARCH (1,1) models are effective in predicting risk across all three periods. However, statistical model comparison indicates that the GJR-GARCH (1,1) model outperforms the GARCH (1,1) model consistently across all periods. Nevertheless, its advantage is most pronounced during the COVID-19 period, when extreme market turbulence and asymmetric volatility were present. These results support the necessity of solid volatility modeling regarding REIT risk management, particularly in emerging markets and under both normal and extreme market conditions. This study makes a novel contribution by being the first to apply and compare GARCH-type models specifically to the Saudi REIT market across these pandemic-defined subperiods. It addresses a gap in regional volatility modeling and demonstrates the superior performance of asymmetric models under crisis conditions, offering insights for REIT risk management in emerging markets.