Frank copula on value at risk (VaR) of the construction of bivariate portfolio (Case Study: stocks of companies awarded with the IDX top ten blue with stock period of 20 october 2014 to 28 february 2018)

Abstract

Value at Risk (VaR) is a method to estimate the worst risk of an investment. The stock data is one of the financial time series data which often has high volatility which causes inconstant residual variance. The combination between several stocks in the portfolio makes the assumption of residual normality of the joint distribution model difficult to fulfill. The previous research on VaR by Sofiana in 2011 [3] and Hermansyah in 2017 [4] found that VaR value was reliable only for the data fulfilling normality assumption. Therefore, it is necessary to estimate VaR without ignoring the presence of heteroscedasticity and unfulfilled residual normality of the joint distribution model. This research aims to measure the VaR using Frank Copula-GARCH method with stock return data of BBRI, TLKM and UNVR for the period of 20 October 2014 to 28 February 2018. The research found that a pair of bivariate portfolio was TLKM and UNVR because they had the highest residual correlation value of Rho Spearman of ρ = 0.3204. Based on the data generation obtained using Monte Carlo simulation, the results of the VaR were -0.027883; -0.01886425; -0.01403 with confidence level at 99%, 95%, and 90% respectively.