Introduction and Performance Comparison of some Common Multi-period VaR Forecasting Methods: A Case Study of the Tehran Stock Exchange

Abstract

According to the Basel accords, financial institutions should forecast VaR of their portfolio over multi-period time horizons in order to determine their capital adequacy. Hence, finding efficient models for forecasting multi-period VaR is crucial for Chief Risk Officers (CRO) in general and Financial Risk Mangers (FRM) in particular. This paper tries to analyze and compare the predictive power of parametric, nonparametric and semiparametric methods of forecasting multi-period VaR for portfolios invested in Tehran Stock Exchange. The results indicate that the “square-root-of-time rule” (the conventional approach) in majority of time horizons has a weak prediction ability vis-a-vis the other multi-period VaR models. Empirical results suggest, employing parametric methods results in larger losses and lower opportunity costs relative to nonparametric methods.

Keywords

• multi-period value-at-risk
• multi-period volatility forecasting
• backtesting