Abstract:

Market risk and uncertainty are the most fundamental risk factors that need to be estimated for investors' decisions. One of the most widely used risk measurement methods is the value-at-risk (VaR) estimation. In this study, the Value-at-Risk (VaR) approach was used for market risk estimation by obtaining the daily gold price return series for the period 02.01.2019-22.08.2022 covering the Covid-19 process in Turkey. The ARMA(1,1)-FIEGARCH(1,d,1) and ARMA(1,1)-FIAPARCH(1,d,1) models, which take into account the long memory and asymmetry characteristics in the return volatility of gold prices, was estimated with different distributions (Normal, Student -t GED and Skewed Student-t). According to the model estimation results, the existence of long memory and asymmetric effects in the volatility of the gold return series were found to be statistically significant. According to the model selection criteria, ARMA(1,1)-FIEGARCH(1,d,1) with the Skewed Student-t Distribution and ARMA(1,1)-FIAPARCH(1,d,1) with Normal Distribution has been estimated as suitable models for the return series volatility. Value-at-risk (VaR) estimates for gold market risk are obtained in short and long positions based on both models. The study findings support that value-at-risk (VaR) estimates based on volatility models that take long memory and asymmetric effects into account are appropriate for gold market risk.

Keywords: