Forecasting Volatility for an Optimal Portfolio with Stylized Facts Using Copulas
- PDF / 1,819,889 Bytes
- 22 Pages / 439.37 x 666.142 pts Page_size
- 64 Downloads / 195 Views
Forecasting Volatility for an Optimal Portfolio with Stylized Facts Using Copulas Aida Karmous1 · Heni Boubaker1,2 · Lotfi Belkacem1 Accepted: 14 August 2020 © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract In this paper, we seek to examine the effect of the presence of stylized facts on forecasting volatility and we model the dependence between exchange rate returns using a flexible approach that allows us to investigate whether the co-movement of the different stylized facts on portfolio optimization. First,we focus on the dependence structure using copulas. The empirical results show that the co-jumps, long memory, leverage effects affect the dependence structure. Second, we analyze the impact of the presence of stylized facts with the dependence structure using Gumbel copula on the optimal portfolio. We propose a new approach to forecasting volatility portfolio with dynamic factor models including stylized facts and assuming that the dependence structure is modeled by the copula parameter. The empirical results show that our approach outperforms the basic models without stylized facts and where the dependence structure is represented by the linear correlation coefficient. Keywords Dynamic factor model · Multivariate stochastic volatility · Co-jumps · Leverage · Long memory · Copulas model · Portfolio optimization JEL Classification C14 · C32 · G11 · G15
1 Introduction Many researches are interested by understanding the co-movements between financial assets, index or exchange rate for constructing an optimal portfolio diversification. Researchers proved that co-volatility are the source of co-movement between international equity markets as like Solnik et al. (1996) and Andersen et al. (2001b). * Heni Boubaker [email protected] 1
IHEC of Sousse, Research Laboratory for Economy, Management and Quantitative Finance (LaREMFiQ), B.P. 40, 4054 Sousse, Tunisia
2
Rabat Business School, BEAR LAB (UIR), Technopolis Rabat-Shore, 11100 Rabat‑Salé, Morocco
13
Vol.:(0123456789)
A. Karmous et al.
In the financial community, it is understood that asset prices do not evolve as prescribed by the efficient market hypothesis which various statistical properties of asset returns are then described: asymmetric, distributional properties, tail properties and extreme fluctuations, path wise regularity, linear and nonlinear dependence of returns in time and across stocks. It is more common in the econometrics research and literature to use daily, monthly or yearly data; these kinds of data do not take into account the totality of the information. In other words, these data give allowance for overlooking instantaneous alterations that in a minute can cause crisis and other kinds of risks. Today high frequency financial data are becoming increasingly used in the most research in financial econometrics where information with high frequency can capture many stylized facts and we exploit for different purposes such as estimation and forecasting asset prices, exchange rate and portfolio. For
Data Loading...
