A1983
Title: Stochastic correlation modelling with Von Mises process
Authors: Sourav Majumdar - Indian Institute of Management Ahmedabad (India) [presenting]
Abstract: Multi-asset financial derivatives can carry an intra-asset correlation risk. Previous empirical studies show that the correlation between assets is not stable, and there is a correlation risk premium. Pricing models for these derivatives should account for the time-varying nature of the correlation. Assuming constant correlation may lead to pricing and hedging risks. We propose a continuous-time model for correlation as a random variable on the circle. We consider a model based on the von Mises process whose stationary distribution is the von Mises distribution, the maximum entropy distribution on the circle. We discuss several estimation methods and results for the model. We apply the model to real-life financial data to study the correlation between equity and currency exchange rates.
